The reaction time-based Concealed Information Test (RT-CIT) is a memory task that detects one’s reaction time cost when viewing a familiar photograph. Thus, the guilty should react slower to crime-related pictures relative to pictures not from the crime. The RT-CIT, however, relies heavily on suspects’ memory integrity and is therefore potentially susceptible to factors that are known to impair memory integrity or interfere with memory-matching capability. Here, we tested one possible weakness of the RT-CIT: namely, how accurate (and resemblant to suspects’ memory) do the relevant pictures have to be? Our participants saw only one viewpoint of a probe object, but were tested with images of 12 different angles of the same probe item. To our surprise, we found the RT-CIT to be quite resistant to picture rotations across all angles, and this was true for both deep-encoding (Experiment 1) and shallow-encoding (Experiment 2) conditions. Furthermore, for deep-encoders, the RT-CIT was most effective under 250 trials, but remained accurate within almost 800 trials. We conclude that, for well-remembered items or pictures, the forensic team does not need to worry about reconstructing the angles (e.g., approximate the eye height and tracing the footsteps of the perpetrator) that closely match what the perpetrators saw.