We investigated sources of the asymmetric bias found in estimation of a time interval (250-400 ms) embedded in the Space Fortress task (Donchin, 1989). Two hypotheses to explain this bias were tested in a behavioral experiment: 1) contamination from a different time interval representation, and 2) pressure to complete the task in time. Participants alternated between producing the target interval and producing either a shorter or a longer interval while the total time allowed for the task was manipulated. The results showed that the target interval estimate was significantly influenced by both manipulations. The effects were captured by incorporating the timing model of Taatgen and Van Rijn (2011) into the ACT-R model for Space Fortress (Bothell, 2010). Time estimation performed in a dynamic task requires understanding the influence of external temporal tasks as well as the procedural demands of performing multiple tasks under time pressure.