OCTOBER 10, 2017 STB UPDATE

OCTOBER 10, 2017 STB UPDATE

Network performance recovered and continuing to improve Dwell lower than 2016, as well as Q1 and Q2 2017 (averages); velocity performance at best level since June, with both velocity and transit on similar trend to April 2017 s strong progression Right Car Right Train stable Crew and power resource levels remain well matched to demand Hump yards performing reliably and efficiently Western terminals recovered, secondary congestion recovered as well Empty car fulfillment stable, demand down modestly following month-end surge Local pull and place performance stable Customer problem logs have returned to normal levels as performance has improved Interchange volumes and performance steady Significant operating improvements during third quarter put CSX on a solid foundation for fall peak 2

9.9 10.1 10.2 10.5 10.2 12.5 11.8 15.7 15.8 15.8 16.0 15.9 15.2 15.9 63% 65% 66% 76% 76% 70% 79% 77% 76% 70% 73% 74% 70% 71% Dwell and velocity improved, performance at best levels since June On Time Originations (%) On Time Arrivals (%) 100% 80% 60% 86% Weekly Average 76% 100% 80% 60% 75% 68% On-Time +2 hrs 40% 40% On-Time 20% 0% 20% 0% Sep. 30 Oct. 6 Sep. 30 Oct. 6 +2 hrs Dwell (hours) Velocity (mph) 16.0 18.0 14.0 12.0 10.0 11.8 11.2 16.0 14.0 15.6 15.1 8.0 12.0 6.0 10.0 Sep. 30 Oct. 6 Sep. 30 Oct. 6 3 Note: Dwell and velocity displayed according to CSX methodology; explanation of CSX methodology can be found in appendix. Week 37 & 38 dwell and velocity exclude terminals that held cars through Hurricane Irma-impacted period and specific trains held through storm, respectively

Right Car Right Train holding stable; less relevant in PSR 85% 80% 75% 70% 75% Right Car Right Train 1 75% 74% 67% Weekly Average 73% 77% 78% 79% 75% Right Car Right Train is no longer a measure that CSX uses to manage its operation In precision scheduled railroading (PSR), if a car can be advanced on another train to speed transit or ensure its on-time arrival, there is not one right train Car priority is to move cars quickly, on next available train 65% Asset utilization a key tenet of PSR 60% 55% Train priority is blocking integrity and departing all available, relevant cars from the yard 50% Sep. 30 Oct. 6 Blocking integrity certifies that a train is built correctly and shipments are headed to the correct location Managed through field supervision 4 1 Right Car Right Train is defined as the percentage of cars that departed from a yard in accordance with their car scheduling trip plan

Resourcing appropriately to meet business needs Active Locomotives Train & Engine Headcount and Re-crew Rate 1 4,000 3,600 3,763 9,900 9,700 10% 9% 8% 3,200 3,182 9,500 7% 6% 2,800 2,400 9,300 9,100 9,223 9,278 5% 4% 3% 2% 2,000 8,900 1% 0% Q3 locomotive level stable; engines in place to support grain harvest season Re-crew rates remain at historic lows and stable Power and crew availability steady in third quarter at approximately 99% and 95%, respectively 5 1 Re-crew rate is re-crew people starts as a percent of total measured people starts, and represents incidences of replacing a crew on the same train ID (generally due to hours of service).

15.2 16.6 17.8 17.3 19.7 18.6 18.3 Hump yard performance steady CSX Hump Terminal Overview Transitioned to flat-switching operations Hump terminals Selkirk, NY Total hump yard volumes remain in a consistent band week-over-week, well below capacity of yards Key hump productivity and efficiency measures performing well Dwell continues to improve, at lowest levels since June Toledo, OH Avon, IN Willard, OH Cumberland, MD Cincinnati, OH Dwell at Hump Terminals 1 Louisville, KY Nashville, TN Birmingham, AL Atlanta, GA Hamlet, NC Waycross, GA 26.0 24.0 22.0 20.0 18.0 19.5 18.9 Weekly Average 16.0 14.0 12.0 Absolute number of humps not good or bad ; goal is best mix of hump and flat yards for processing efficiency 10.0 Sep. 30 Oct. 6 6 1 Dwell displayed according to CSX methodology; explanation of CSX methodology can be found in appendix. Week 37 & 38 dwell excludes terminals that held cars through Hurricane Irma-impacted period

12.4 12.9 12.6 12.5 11.5 11.6 14.7 Western performance recovered; plan changes alleviated congestion Western Corridor Key Terminals Avon, IN Evansville, IN Nashville, TN Key terminal productivity and performance measures recovered in former trouble spots Train plan addressed secondary concerns Leveraged Avon as offset of increased volume flow through Russell, Columbus and Louisville Dwell at these three locations down 51% from high point and an additional 11% week-over-week Dwell at Western Terminals 1 Birmingham, AL Montgomery, AL Mobile, AL Western terminals 24.0 22.0 20.0 18.0 16.0 14.0 12.0 10.0 8.0 6.0 16.0 12.5 Weekly Average Sep. 30 Oct. 6 7 1 Dwell displayed according to CSX methodology; explanation of CSX methodology can be found in appendix. Week 37 & 38 dwell excludes terminals that held cars through Hurricane Irma-impacted period.

Car order fulfillment remained at higher levels Cars Ordered 7,000 6,000 5,000 4,000 3,000 2,000 Weekly Car Orders and Normalized Fill Rate Normalized Fill Rate 90% 85% 80% 75% 70% 65% 60% 55% 50% 45% 40% Car orders down modestly following month-end surge seen in prior week Empty car dwell continues to rise at customer locations Indicates improved car supply and availability Customers maintaining buffer stock, which elongates total asset turn times Weekly Orders Normalized Order Fill % Q1 Avg. Weekly Car Orders 8 1 Normalized fill rate is a proxy of demand fulfillment against historical/expected order levels, as current order levels are disconnected with demand

Last mile performance stable 100% 90% 80% 95% 90% Local Service Measurement 1 84% 84% Local Service Measurement (LSM) is no longer a metric that CSX uses to manage its operation In precision scheduled railroading (PSR), focus on endto-end transit and customer expectations Last mile performance must be in combination with, not independent of, overall performance 70% 60% 50% 40% 30% 20% 10% 0% Q1 AVG Q2 AVG Accordingly, LSM as a reported metric was discontinued upon start of PSR implementation At request of STB, last mile tracking reinstated to monitor through implementation period Data reflects passive information flow, lacking prior focus on field reporting to ensure LSM capture Reliable pull and place expected as part of service to customers 9 1 Local Service Measurement is defined as the percentage of cars that were pulled or placed at a customer location based upon daily customer request, the local service plan and available inventory at the local serving yard

Customer problem logs down further, and in normal range 354 368 286 374 458 Customer Inquiries Daily Average Log Volume 570 563 537 567 499 374 330 281 294 272 265 Delayed cars remain most frequent concern, but have returned to normal levels (<1% of traffic) Trend in problem logs mirrors timeframe of network challenges and recovery, followed by Hurricane Irma Lower levels of long-dwelling cars reflects overall fluidity improvements Nearly 90% of last two weeks problem logs have been addressed and closed to-date Leaving more logs open through final destination Managing pipeline of customer concerns to full resolution Delayed Cars Bad Order Switching Issues 10

Interchanges remain current and performing to expectations 800 East St. Louis Daily Average Interchange Volume From To 2,000 Chicago Daily Average Interchange Volume 600 1,600 400 200 0 1,200 800 400 0 600 New Orleans Daily Average Interchange Volume 300 Memphis Daily Average Interchange Volume 400 200 200 100 0 0 11

Precision scheduled railroading producing service improvement Operational Focus Service Improvements & Productivity Improvements Balanced Train Plan Improved Frequency Better Reliability Rolling Stock Utilization People Efficiency Terminal Fluidity Faster Transit Quicker Turnaround Fuel Optimization Train Density Improve Service Operate Safely Control Costs Drive Asset Utilization Develop People Realigned service frequency in second quarter Set the groundwork of a balanced train plan in early July Terminals improved efficiency and traffic flow adjustments have significantly recovered service Improved execution on this foundation to drive long-term service and productivity improvements 12

APPENDIX

CSX has changed methodology on some metrics reported publicly Velocity Dwell Cars Online Former Line of road miles per hour Former Car time at terminal, excluding cars on the same train ID Former All cars on CSX, as determined by RailInc Current Total miles traveled per hour, including intermediate dwell of the train Current All car time with a terminal work event, including through cars on same train ID (e.g. crew change) Current RailInc cars on CSX, excluding cars stored, under repair, sold, and private cars ex online inventory Change Reason Includes full trip of a train and ability to diagnose overall speed profile (in support of improvement in asset cycle) Change Reason Includes all dwell with ability to diagnose all events impacting car movement (in support of improvement in asset cycle) Change Reason More accurate measurement of active cars on line, i.e. cars for which CSX is focused on realtime, efficient movement Effect on Metric Reported velocity will be lower Effect on Metric Reported dwell will be lower Effect on Metric Reported cars online will be lower Restated historical data in new methodology available on csx.com/servicemetrics 14