Eyewitness lineups are typically composed of a suspect (guilty or innocent) and fillers (known innocents). Meta-analytic techniques were used to investigate the extent to which manipulations of suspect-filler similarity influenced identification decisions. Compared with lineups with moderate or high similarity fillers, lineups with low similarity fillers were far more likely to elicit suspect identifications. This was true regardless of whether the suspect was guilty or innocent, underscoring the importance of ensuring the suspect does not stand out from the fillers. Although whether the lineup contained moderate or high similarity fillers had no reliable influence on guilty suspect identifications, a higher rate of innocent suspect misidentifications was found for moderate similarity lineups. The correspondence between the meta-analytic findings and current lineup construction recommendations is discussed.