Purpose: The purposes of the study were to investigate (a) the task-specific differences in short-term memory (STM) and working memory capacity (WMC) in individuals with mild cognitive impairment (MCI) and normal elderly adults (NEAs), (b) the Stroop interference and facilitation effects, and (c) the relationship of STM and WMC to the Stroop effects. Method: Thirty-two individuals participated in the study (n = 16 for each group). WMC demands were increased using a computerized Stroop-like token task to add more linguistic units. Six STM and WMC measures were administered overall. Results: Digit-related tasks and an alphabet span task sensitively differentiated individuals with MCI from the NEA group. The group with MCI exhibited greater Stroop interference effects than the NEA group, but the 2 groups did not exhibit different Stroop facilitation effects. WMC significantly predicted performance on the response time analyses but not on the error rate analyses. Conclusion: Task-specific differences emerged in the group with MCI, and a reduced WMC accounts for the impaired inhibitory and goal maintenance processes. It is critical that WMC demands be systematically manipulated to tax individuals' WMC in a way that can clearly demonstrate their deficits, especially in individuals who are at risk for clinically demented states. (Contains 3 tables and 4 figures.)