The paper introduces HOTA (Higher Order Tracking Accuracy), a novel evaluation metric for Multi-Object Tracking (MOT) that balances the importance of detection, association, and localization. HOTA decomposes into a family of sub-metrics that evaluate five basic error types separately, providing clear analysis of tracking performance. The metric is evaluated on the MOTChallenge benchmark and shown to capture aspects of MOT performance not previously addressed by established metrics. It aligns better with human visual evaluation of tracking performance compared to existing metrics like MOTA and IDF1. HOTA is designed to be a unified metric for ranking trackers while also allowing for detailed analysis of different tracking aspects. The paper provides a thorough theoretical analysis of HOTA and compares it to other metrics, highlighting its advantages and addressing their shortcomings.The paper introduces HOTA (Higher Order Tracking Accuracy), a novel evaluation metric for Multi-Object Tracking (MOT) that balances the importance of detection, association, and localization. HOTA decomposes into a family of sub-metrics that evaluate five basic error types separately, providing clear analysis of tracking performance. The metric is evaluated on the MOTChallenge benchmark and shown to capture aspects of MOT performance not previously addressed by established metrics. It aligns better with human visual evaluation of tracking performance compared to existing metrics like MOTA and IDF1. HOTA is designed to be a unified metric for ranking trackers while also allowing for detailed analysis of different tracking aspects. The paper provides a thorough theoretical analysis of HOTA and compares it to other metrics, highlighting its advantages and addressing their shortcomings.