For a long time, post-reinforcer delays have been considered to have no effect on choice. The most influential choice models therefore do not consider such delays. Recently, however, some studies reported that post-reinforcer delays affect choice behavior in pigeons. One such study (Mazur, 2006) concluded that a short initial link increases the sensitivity to post-reinforcer delays in concurrent-chains schedules. However, this study did not use typical concurrent-chains schedule procedures, in which the number of reinforcements differs between alternatives, and did not systematically analyze the effect of the systematic post-reinforcer delay. The current study therefore examined whether the length of the initial link modulates the post-reinforcer delay effect on choice with standard concurrent-chains schedules and by systematically varying both the initial link and the post-reinforcer delay. As the results, the shorter the initial link length, the larger the effect of the post-reinforcer delay. Models with post-reinforcer delays predicted choice behavior better than models without, but pigeons were more insensitive to delays than the model prediction for long initial links. These findings provide important information for models on choice, and suggest that experiments should be designed with more caution with regards to post-reinforcer delays.