Development and application of maintainability models have been seriously constrained by the paucity of valid human performance repair time data. Moreover, the high costs of obtaining such data by empirical (observational) methods lend little hope that useful quantities of data will ever be available for widespread use of models. This state-of-affairs suggests the alternative, far less costly, method of acquiring these data by subjective judgments. This study therefore sought to explore the potential capabilities and limitations of highly experienced technicians to provide repair time estimates on 23 electronic equipments. Results indicated high agreeement among technicians across all equipments. Although validation work was not possible within the constraints of the study, the estimates were compared with data reported in the world-wide AF 66-1 maintenance reporting system. These latter data are known to be inflated, thus limiting their usefulness as model inputs. The result of this comparison revealed that (a) all subjectively derived MTTR's were substantially lower than those generated from the AF 66-1 system, and (b) the correlation between the two sets of MTTR's was a remarkably high .856. These findings strongly suggest that the estimated MTTR's may have practical validity; at minimum, they must be considered more representative of actual repair times than those available in the AF 66-1 system. Validation of these estimates and further study of subjective techniques is highly recommended.