Manhattan Institute for Policy Research.
Subscribe   Subscribe   MI on Facebook Find us on Twitter Find us on Instagram      


Issue Brief

No. 23 June 2013

Pipelines Are Safest For Transportation of Oil and Gas

Diana Furchtgott-Roth, Senior Fellow, Manhattan Institute

Manhattan Institute.

Research assistance for this report was provided by Claire Rogers and Andrew Gray.

The Obama administration’s decision to delay approval for the construction of TransCanada Inc.’s proposed Keystone XL pipeline was based, in part, on concerns over the safety and reliability of oil and natural gas pipelines. The pipeline is intended to transport oil from Canada to U.S. refiners on the Gulf of Mexico. In announcing his decision, the president called for a full assessment of “the pipeline’s impact, especially on the health and safety of the American people.”

Pipelines have been used to transport American natural gas or oil, including from Canada to the United States, for three quarters of a century. Almost 500,000 miles of interstate pipeline crisscross America, carrying crude oil, petroleum products, and natural gas. This extensive and operational infrastructure network is heavily regulated by the Department of Transportation, which monitors the very issues central to the Keystone controversy: safety and reliability.

Thus it is possible to answer, based on experience, the question of whether pipeline transport of oil and gas is safe. It is, moreover, possible to compare the record of oil and gas pipelines to that of transport via rail and road. As the major alternative means of fuel shipment, transport by rail and road has been increasing as limitations on pipeline capacity have become manifest (the underlying reason for the Keystone proposal).

A review of safety and accident statistics provided by the U.S. Department of Transportation for the extensive network of existing U.S. pipelines—including many linked to Canada—clearly show that, in addition to enjoying a substantial cost advantage, pipelines result in fewer spillage incidents and personal injuries than road and rail. Americans are more likely to get struck by lightning than to be killed in a pipeline accident.[1]

The question of how to transport oil and gas safely and reliably is not a transitory one linked only to the Keystone controversy. Petroleum production in North America is now nearly 18 million barrels a day,[2] and could climb to 27 million barrels a day by 2020. Natural gas production in Canada and the United States could rise by a third over the same period, climbing to 22 billion cubic feet per day. This oil and gas will have to travel to where it is needed. Whether it is produced in Canada, Alaska, North Dakota, or the Gulf of Mexico, it will be used all over the country, especially since new environmental regulations are resulting in the rapid closures of coal-fired power plants, increasing the demand for natural gas as a substitute. Similarly, large fleets of buses and trucks are switching to natural gas, and General Motors and Chrysler are making dual-fuel pickup trucks.

This paper compares the record of transport via pipeline to that of road and rail and finds that pipelines are the safer option.

The first large-diameter long-distance pipelines were constructed during the Second World War, and they proliferated across the country over the ensuing two decades. Now America has 175,000 miles of onshore and offshore petroleum pipeline and 321,000 miles of natural gas transmission and gathering pipeline. In addition, over 2 million miles of natural gas distribution pipeline send natural gas to businesses and consumers.[3] This is expected to increase as households and businesses shift to natural gas to take advantage of low prices that are expected to last into the foreseeable future.

Pipelines are the primary mode of transportation for crude oil, petroleum products, and natural gas. As shown in Table 1, approximately 70 percent of crude oil and petroleum products are shipped by pipeline on a ton-mile basis. Tanker and barge traffic accounts for 23 percent of oil shipments. Trucking accounts for 4 percent of shipments, and rail for the remaining 3 percent. Essentially all dry natural gas is shipped by pipeline to end users.

If safety and environmental damages in the transportation of oil and gas were proportionate to the volume of shipments, one would expect the vast majority of damages to occur on pipelines. This paper finds the exact opposite. The majority of incidents occur on road and rail.

Data on pipeline safety are available from the United States Department of Transportation Pipeline and Hazardous Materials Safety Administration Office of Pipeline Safety (PHMSA).[4] Operators report to PHMSA any incident that crosses a certain safety threshold. These reports enable the public to compare the safety of pipelines to that of road and rail.

A pipeline incident must be reported if any of the following occur: (1) Explosion or fire not intentionally set by the operator; (2) Release of five gallons or more of a hazardous liquid (any petroleum or petroleum product) or carbon dioxide; (3) Fatality; (4) Personal injury necessitating hospitalization; and (5) Property damage, including cleanup costs, and the value of lost product, and the damage to the property of the operator or others, or both, estimated to exceed $50,000.[5]

One way to look at the safety record of petroleum, petroleum products, and natural gas pipeline operators is to examine PHMSA’s aggregated data from individual reports. Table 2 shows a summary of all reported incidents and damage between 1992 and 2011. Property damage costs are reported by PHMSA in 2011, with lost product accounted for at benchmark prices at the time of the incident.

To the untutored eye, it can appear that pipelines are prone to significant accidents. For instance, there were 721 incidents in 2005, and 53 fatalities in 1996, many caused by a propane explosion in San Juan. However, as the tables make clear, safety-related incidents, as measured by volume, are actually minor. More importantly, it is crucial to keep in mind that there is no way, in an advanced industrial economy, to avoid shipment of fuels to provide power. Crucially, by comparison with other means of such transport, pipelines emerge as relatively safe and reliable.

Table 2 shows that the number of incidents is relatively low. It has ranged from 339 in 1999 to 721 in 2005. Property damage has ranged from $53 million in 1995 to $1.3 billion in 2010. Lost barrels of liquids reached a low of 32,258 barrels in 2009 to a high of 123,419 the following year. Injuries ranged from 36 in 2006 to 127 in 1996, and fatalities ranged from 7 in 2001 to 53 in 1996.

The unusual increases in gross property damage in 2005 and 2010 were largely attributable to Hurricane Katrina in 2005 and the Kalamazoo River oil spill in 2010. Higher market prices for petroleum over the period has led to an increased valuation of spillage. Throughout the 1990s, apart from a brief price spike associated with the Persian Gulf War, the West Texas Intermediate wholesale price of oil stayed below $25 dollars per barrel. Prices continued to increase between 2000 and 2008, and averaged $100 in 2008. Prices eased in 2009 and 2010, but averaged around $95 in 2011 and $94 in 2012.[6]

A major criterion for determining if an incident had to be reported to PHMSA was significantly revised in 2002. Between 1992 and 2002 a spill only had to be reported if it was greater than 50 barrels of liquids or CO2 (after 1991). However, beginning in 2002, the limit was dropped to five gallons, with an exception for maintenance-related spills of five barrels or less confined to company sites.[7] Hence, minor spills that were not reported prior to 2002 were reported afterwards. From 1992 through 2001 an annual average of 383 incident reports were filed with PHMSA. Then, from 2002 through 2011, companies filed an annual average of 644 incident reports.

Gross barrels spilled do not take into account the number of barrels that were recovered during cleanup. The volume of liquids spilled that is ultimately recovered varies widely from year to year, and is likely heavily influenced by the nature of the spill. Between 1992 and 2011 about 40 percent of spilled liquids were recovered (Table 3). Over the entire 20-year period a total of less than 1.5 million net barrels were spilled.

Volumes that are spilled are miniscule when compared to the volumes of petroleum that are used in the United States. To provide some prospective, U.S. refineries produce over 7 million barrels of gasoline every single day.[8] Considering the vast network, 175,000 miles of petroleum pipeline and over 2 million miles of natural gas pipelines (about 321,000 of transmission and gathering lines, over 2 million of local distribution main and service lines), incidents are exceedingly rare.[9]

To draw another comparison, according to the National Weather Service there was an average of 35 reported deaths annually caused by lightning from 2003 to 2012.[10] From 1992 to 2011 fatalities related to pipeline incidents were about 20 per year. An individual had a 75 percent greater chance of getting killed by lightning as being killed in a pipeline incident.

Data are also provided by PHMSA that make it possible to determine in what type of pipeline system a particular incident occurred. There are four basic categories of pipeline systems, namely hazardous liquids, natural gas gathering, natural gas transmission, and natural gas distribution. Natural gas gathering pipelines bring raw natural gas from the wellhead to the gas processing plant. The natural gas transmission system is made up of pipelines that bring processed (dry) gas from the plants and carry it across the country to city gates or to large customers (e.g., heavy industry or electrical power plants). The natural gas distribution system is operated by local distribution companies which transport gas from the city gate to local households and local businesses. Table 4 displays what percentage of incidents, fatalities, injuries, and property damage from 1992 through 2011 occurred in each pipeline system.

Although fatalities and injuries are relatively low, the majority of those that do occur have been associated with pipelines that are part of a natural gas distribution system. The U.S. natural gas distribution pipeline network spans over 2 million miles, and the federal government does not regulate intrastate pipelines (local distribution and production gathering lines), except for gathering lines that are located on federal lands. Local distribution companies, where both the vast majority of pipeline miles exist and accidents occur, are regulated by states and municipalities.

The proportion of property damage from incidents originating at hazardous liquids pipelines is largely the result of the inclusion of lost product as part of the damage, and that cleanup of oil spills is costly. From an operational standpoint, incidents associated with natural gas transmission and hazardous liquid systems (large diameter interstate pipelines) have resulted in 86 deaths and 387 injuries from 1992 through 2011, as shown in Table 5.

How does this compare with road and rail? We have analyzed U.S. Department of Transportation data and produced incident and injury rates for oil and gas pipelines, road, and rail for petroleum products in the period 2005 through 2009.[11] Because reporting of pipeline incidents is only required for events involving injury or release over 5 gallons, we eliminated road and rail incidents not meeting those criteria from consideration. Even after this narrowing of scope, road and rail have higher rates of serious incidents and injuries than pipelines, even though more road and rail incidents go unreported.

Table 6 compares incident rates for road, rail, oil and petroleum products pipelines, and natural gas transmission. Road had the highest rate of incidents, with 19.95 per billion ton miles per year. This was followed by rail, with 2.08 per billion ton miles per year. Natural gas transmission came next, with 0.89 per billion ton miles. Hazardous liquid pipelines were the safest, with 0.58 serious incidents per billion ton miles.

Data in Table 7 include all hazmat, not just petroleum products. With respect to pipeline systems, natural gas transmission lines had the lowest average fatality rate for operator personnel and the general public between 2005 and 2009, with a rate of one person killed per year. This was followed by oil and rail, each with an average of 2.4 people per year. The rail figure is skewed by a chlorine incident on January 6, 2005 in Graniteville, South Carolina. Since chlorine is not transported by pipeline, some question the validity of drawing such a comparison. The highest fatality rate is road, with an average of 10.2 people a year. This is not because members of the public are killed due to road accidents with oil trucks. Only 1.4 members of the public, on average, were killed annually, but an average of 8.8 operators died per year.

As shown in Table 8, rates of injury requiring hospitalization and of injury in general show a similar pattern. On average, annual injuries for 2005 through 2009 were lowest for hazardous liquid pipeline, at 4 people with injuries requiring hospitalization per year. The rate was higher for rail, at 4.6 of such injuries per year, although for rail this number was heavily biased by the 2005 observation. Road accidents hospitalized 8.8 people per year, and natural gas pipelines hospitalized 45 people each year.

The rates of injury per ton-mile in Table 8 are most pertinent, however. On this measure, hazardous liquid pipelines outperformed rail and road by a wide margin, causing just 0.0068 injuries requiring hospitalization per billion ton-miles. Rail caused nearly 30 times that many injuries requiring hospitalization on a per-ton-mile basis. Rail was also outperformed by natural gas pipelines on this measure, causing over 1.4 times as many serious injuries per ton-mile. Road was the worst performer on this measure, averaging one quarter serious injuries per billion ton-miles. This is 37 times the hazardous liquid pipeline rate.

Some claim that pipelines carrying Canadian oil sands crude, known as diluted bitumen, have more internal corrosion, and are subject to more incidents.[12] However, PHMSA data show that oil releases from corrosion are no more common in pipelines carrying Canadian diluted bitumen than in other lines.[13] Oil sands crude has been transported in American pipelines for the past decade.

The evidence is clear: transporting oil and natural gas by pipeline is safe. Furthermore, pipeline transportation is safer than transportation by road, rail, or barge, as measured by incidents, injuries, and fatalities—even though more road and rail incidents go unreported.[14]

Despite their safety, pipelines release more oil per spill than rail—but less than road. As Table 9 shows, typical release volumes on rail, particularly of petroleum products, are relatively low at 3,504 gallons per billion ton-miles. While it outperforms road in terms of product release per ton-mile, pipeline transport of petroleum products still experienced product release of 11,286 gallons per billion ton-miles. This figure does decrease by approximately one third if the high product-recovery rate for pipelines is considered, however. Volume release data are unavailable in the PHMSA incident database for natural gas transmission pipelines.

Rising oil and natural gas production is outpacing the transportation capacity of our inadequate national pipeline infrastructure. The Association of American Railroads reports that between 2008 and 2011 the total share of oil and gas rail shipments grew dramatically, from 2 percent of all carloads to 11 percent.[15] In 2011 alone, rail capacity in the Bakken area—stretching from southern Alberta to the northern U.S. Great Plains—tripled to almost 300,000 barrels per day.[16] Crude oil shipments via rail have continued to expand at an accelerating rate; as of September 2012, U.S. Class I railroads were on pace to deliver 200,000 carloads of crude for the year, compared to just 66,000 in 2011 and 9,500 in 2008.[17]

As America continues to ramp up production of oil and natural gas, our pipeline infrastructure becomes more important. We need better pipelines to get oil from North Dakota to the refineries in the Gulf, and natural gas from the Marcellus Shale in Pennsylvania (and New York, should the Empire State allow production to move forward) and the Utica Shale in Ohio to the rest of the country.

In the next few years, the Obama administration may allow more states to explore for oil offshore. In addition, Congress might vote to give coastal areas a share of oil drilling revenue, providing a powerful incentive for more drilling. Congress could also form a liability risk pool to allow independent drillers to expand into the Gulf of Mexico. In order for these resources to get where they are needed, America needs more pipelines—the safest way to move fuel.


This paper contains four major changes in methodology from the prior version of the paper, published in May 2012.

  1. Rail companies are required to report all hazmat releases, but pipeline incidents only require a report if at least 5 gallons of the material are leaked or there is an injury. In the May 2012 version, we reported all hazmat releases. In the revised version, we removed all sub-5-gallon, no-injury road and rail incidents.
  2. We note that one January 6, 2005 chlorine incident skews the rail hazmat fatality statistics. Excluding the chlorine accident would reduce the rail hazmat fatalities in the period we considered from 12 to 3.
  3. We used only petroleum and natural gas product incidents for Tables 6 and 8.
  4. We included a new table (Table 9) showing spillage per ton mile by mode of transportation, also using only petroleum and natural gas products.


  1. Reliable data on water-borne spills, which fall under the jurisdiction of the Coast Guard, are not readily available and so will not be included in this Issue Brief
  2. International Energy Statistics, “Petroleum Production,” accessed May 7, 2013,
  3. “Pipeline Basics,” The United States Department of Transportation Pipeline and Hazardous Materials Safety Administration Office of Pipeline Safety, accessed May 7, 2013,
  4. “Community Toolbox,” The United States Department of Transportation Pipeline and Hazardous Materials Safety Administration Office of Pipeline Safety, accessed April 24, 2012,
  5. “Reporting Criteria Changes,” The United States Department of Transportation Pipeline and Hazardous Materials Safety Administration Office of Pipeline Safety, accessed April 24, 2012,
  6. “Petroleum and Other Liquids, Spot Prices,” United States Energy Information Agency, accessed May 7, 2013,
  7. “Reporting Criteria Changes 1990-Current,” The United States Department of Transportation Pipeline and Hazardous Materials Safety Administration Office of Pipeline Safety, last updated March 2011,
  8. U.S. Energy Information Agency, This Week In Petroleum, accessed May 20, 2013,
  9. “Pipeline Basics,” The United States Department of Transportation Pipeline and Hazardous Materials Safety Administration Office of Pipeline Safety, accessed April 24, 2012,
  10. National Weather Service, “Weather Fatalities,” 2012,
  11. U.S. Department of Transportation, Pipeline and Hazardous Materials Safety Administration, Office of Pipeline Safety, Building Safe Communities: Pipeline Risk and its Application to Local Development Decisions, October, 2010,
  12. See, for example, Lara Skinner and Sean Sweeney, “The Impact of Tar Sands Pipeline Spills on Employment and the Economy,” Cornell University Global Labor Institute, March, 2012
  13. U.S. Department of Transportation, Pipeline and Hazardous Materials Safety Administration, data from form PHMSA F 7000-1
  14. Committee on Transportation and Infrastructure hearings on “Concerns with Hazardous Materials Safety in the U.S.: Is PHMSA Performing Its Mission?” (written report submitted by Majority Staff to the Members of the Committee), September 9, 2009
  15. “EIA: Rail Delivery of Crude Oil and Petroleum Products Rising,” Crude Oil Trader, accessed June 05, 2012,
  16. “Buffet’s Burlington Northern Among Pipeline Winners,” Bloomberg News, accessed June 05, 2012,
  17. Association of American Railroads, “Moving Crude Petroleum by Rail,” December 2012,

The Manhattan Institute, a 501(c)(3), is a think tank whose mission is to develop and disseminate new ideas
that foster greater economic choice and individual responsibility.

Copyright © 2015 Manhattan Institute for Policy Research, Inc. All rights reserved.

52 Vanderbilt Avenue, New York, N.Y. 10017
phone (212) 599-7000 / fax (212) 599-3494