The effect of task-irrelevant numerical values on perceived duration is well established. More precisely, higher numerical values (e.g., ‘9’) correspond to longer estimated durations than lower numerical values (e.g., ‘1’). So far, sparse evidence for two moderators, physical context and stimulus salience, has been provided. The contextual effect refers to an increased difference between estimated durations for low and high numerical values when Arabic digits are presented simultaneously with large physical quantities (e.g., ‘kg’), instead of small physical quantities (e.g., ‘g’). Similarly, the salience effect refers to the observation that differences in time estimations increase when attention is directed to numerical values’ magnitude. Using a time reproduction paradigm, we conducted four experiments to further investigate these two moderators and their possible interaction. In Experiments 1a and 1b, target intervals differed in duration (800, 1000, 1200 ms), numerical value (1, 2, 8, 9), and physical quantity (mg, kg, without). Experiments 2 and 3 additionally included the manipulation of the attentional focus (numerical value or physical quantity) and further quantities (cm, km). Our results supported the positive effect of numerical values on reproduced durations. This was also true for the moderating effect of stimulus salience, which was always significant. In contrast, no evidence for a contextual effect was observed even when participants’ attention was directed on the difference in physical quantity. In conclusion, our data challenge the existence of a moderating contextual effect, while supporting the moderating effect of stimulus salience.