Adaptive control techniques have been applied to the problem of diesel engine torque control. Adaptive control has the potential of greater versatility than classical control techniques. Three adaptive control strategies are tested and compared to each other: self-tuning control with one-shot parameter identification and controller design, self-tuning gain-scheduling control, and self-tuning control with continuous adaptation of system and controller parameters. A continuous-time parameter identification approach, namely the Poisson Moment Functional (PMF) method, is employed due to its superior noise rejection capability. To ensure the applicability of time delay systems, a Smith predictor is employed. The controller design is implemented using a new pole-zero placement algorithm to ensure closed-loop stability. Comparisons with constant parameter controllers reveal that adaptive control provides equal or better torque control than a constant parameter controller. The results of the transient cycle tests also prove that the self-tuning control can be successfully applied to systems with dramatically different dynamic characteristics, and hence show the versatility of the self-tuning adaptive control.