A standard method used to compare database performance involves running searches across databases to calculate the precision of results in each one. The precision of a database search is the number of relevant results obtained for the search topic, divided by the total number of results found. Rating the relevancy of each search result to be able to calculate precision is labour-intensive when the results set is large. This poster will compare the reliability of various precision cut-off levels to compare database performance, using 25 topic searches run in STEM databases. Precision will be calculated for the entire results set, then for the first 10 results listed in each database, the first 20, and in increasing increments of 10 until 100. The purpose of this investigation is to determine whether calculating the precision of a subset of search results provides the same answer, for which is the better performing database, as calculating the precision of the entire results set.