This paper investigates the usage of strategies in the Time-Left task (Gibbon & Church, 1981). In that task, participants are assumed to compare temporal intervals on their subjective timescales (i.e., do temporal arithmetic), yielding different hypotheses for linear and nonlinear subjective time. Here we present an experiment and ACT-R model simulations that show that participants probably use strategies different from temporal arithmetic. Usage of other, alternative strategies would allow for any subjective timescale. As the interpretation of Time-Left results critically depends on temporal arithmetic, these results invalidate the Time-Left task for distinguishing between different internal timescales.