Objectives: Designing the Effective interventions for modifying the risky driving behaviors needs to know about audiences' attitudes and behaviors. In this study, we measured the attitudes, self-reported and observational behaviors among taxi drivers, a large group of professional drivers, in Tehran, the capital of Iran.Methods: In this cross-sectional study, we selected two municipality district of Tehran which had the highest rate of traffic accidents among taxicabs. By using a questionnaire, we asked them about their attitudes and reports of their behaviors during driving and a checklist was used for observing their risky driving behaviors. Then, collected data were analysed.Results: We found that taxi drivers in Tehran had had desired attitudes toward risky driving behavior and had reported their driving behavior well and non- risky. There was a significant direct relationship between their attitudes and driving history. However, we found a reverse significant relationship between self-reported behaviors and literacy level. Risky driving behavior wrer different based on the time of observation. Using cell phone, tailgaiting and not moving between lines were three most common observed risky driving behaviors among taxi drivers.Conclusion: Taxi drivers, as professional drivers which form the large segment of our interventions for modifying the risky driving behaviors, have not desired behaviors regardless of their good attitudes and self reports. Thus, we should design and implement the interventions based on their driving history, literacy level, kind of the common behaviors and even the time of driving.