Similarity models provide an alternative approach to intertemporal choice. Instead of calculating an overall value for options, decision makers compare the similarity of option attributes and make a decision based on similarity. Similarity judgments for reward amounts and time delays depend on both the numerical difference (x2-x1) and ratio (x1/x2) of quantitative values. Changing units of these attribute values (e.g., days vs. weeks) can alter the numerical difference while maintaining the ratio. For example, framing a pair of delays in the unit of weeks (1 vs. 2) or days (7 vs. 14) both result in a ratio of 1/2. Yet the numerical difference between the delays differs depending on the unit (1 for weeks and 7 for days). Here we had participants make similarity judgments and intertemporal choices with amounts framed as dollars or cents and delays framed as days or weeks. We predicted that they units of amounts and delays would influence similarity judgments which would then influence intertemporal choices. We found that participants judged amounts framed as cents as less similar than dollars, and this resulted in more patient intertemporal choices. Additionally, they judged delays framed as weeks as more similar than days, but the framing did not influence choice. These findings suggest that the units in which amounts and delays are framed can influence their similarity judgments, which can shape intertemporal choices. These unit effects may guide stakeholders in framing aspects of intertemporal choices in different units to nudge decision makers into either more impulsive or patient choice.