Aims: The objective of this study was to identify operators' error in distillation units of Isfahan oil refinery. Materials and Methods: Data were collected through task observation and interviewing with safety authorities, the unit and the shift supervisors and operators to identify and analyze critical tasks hierarchically (hierarchical task analysis). Then, human errors of each critical task were identified using systematic human error reduction prediction approach (SHERPA) technique. Results: Analysis of the SHERPA work sheets revealed 198 human errors of which 134 (67.64%), 23 (11.61%), 11 (5.6%), 24 (12.12%), and 6 (3.03%) were action, checking, communication, retrieval, and selection errors, respectively. Critical tasks of "performance monitoring" and "communication" were the main tasks of control room operators (C.R.O's). Low occurrence probability and medium occurrence probability were estimated 64% and 36%, respectively. Furthermore, 59% of the identified errors of C.R.O's had no required recovery of which only 29% had critical consequences. Conclusions: The results showed SHERPA technique can be used as an effective technique to detect human errors in petrochemical and oil refineries.