Two experiments investigated Estes and Maddox’ theory (2002) that word frequency mirror effect in episodic recognition memory is due to word likeness rather than frequency of experience with a word. In Experiment 1, sixteen first year psychology students at the University of Newcastle studied lists of high and low frequency words crossed with high-neighbourhood-density and low-neighbourhood-density words and were given an episodic recognition test and asked to rate words as new or old and provide ratings of confidence according to a three point scale with six possible responses: sure old, probably old, possibly old, possibly new, probably new and sure new. Experiment 2 included twenty-three first year psychology students at the University of Newcastle who were tested using lexical decision task lists of words and nonwords. Testing was undertaken on a computer that presented the stimuli and recorded the participants’ responses using a program written in Turbo Pascal 6.0 with millisecond accurate timing. The dataset contains one Microsoft Excel file in .xls format containing data for Experiments 1 and 2.