Neuropsychological tests have become commonplace in the assessment of sports-related concussion. Typically, post-injury test data are compared to pre-injury "baselines." Baseline testing can be expensive and logistically challenging, yet the usefulness of neuropsychological baseline testing has not been tested empirically. This paper examines the extent to which baseline testing is useful for detecting neurocognitive deficits following sports concussion in a college-age population. A total of 223 collegiate athletes from multiple sports who sustained concussions and had both baseline and post-injury testing using Immediate Post-concussion Assessment and Cognitive Testing (ImPACT) were included in the study. Reliable change (RC) in scores was determined by two approaches, the Jacobson and Truax (JT) and the Gulliksen-Lord-Novick (GLN) methods. The 90% confidence interval was used for both. Classification using these methods was compared to standard normative methods that compared post-concussion performance to baseline population means. Agreement between reliable change and normative methods was examined using Cohen's Kappa scores to determine whether post-injury scores alone could identify reliable cognitive decline. Mean time from concussion to post-injury testing was 3.40 days. The percentage of athletes who declined when using the JT method was similar to the percentage that would be expected to decline due to chance alone. Although the GLN and JT methods demonstrated moderate to substantial agreement, the GLN method consistently identified more cognitively compromised athletes than the JT method. Post-injury scores alone identified a significant majority of athletes with a reliable decline on ImPACT. Although preliminary and in need of replication across age groups and instruments, these findings suggest that the majority of collegiate athletes who experience clinically meaningful post-concussion cognitive decline can be identified without baseline data.