The ability to discriminate between physical magnitudes (e.g., loudness, brightness, duration) is a basic component of human judgment. Since the time of G.T. Fechner, discrimination performance has often been studied with the 2-Alternative-Forced-Choice (2AFC) paradigm, wherein which participants repeatedly compare a constant standard stimulus against a variable comparison stimulus. For the case of duration discrimination, it has been shown repeatedly that discrimination performance is better when the standard precedes rather than follows the comparison, a phenomenon which is referred to as the Type-B-Effect (TBE). This effect is not only counterintuitive but also contradicts standard psychophysical models such as Signal Detection Theory. Here, we present the results of a meta-analytic random-effects model designed to assess the TBE’s generality and size. This meta-analytic regression model indicated real evidential value for the TBE and thus reveals it as a ubiquitous feature of the classic 2AFC task. Accordingly, future models of stimulus discrimination should ideally account for this effect. Although the mechanisms responsible for the occurrence of the TBE are currently not fully understood, memory updating, internal reference formation, and differential weighting of the two stimulus positions are plausible candidate mechanisms that are in line with our findings.