The suspect's camouflage garment and the garment worn by the perpetrator captured by a surveillance camera can be matched only if specific unique characteristic of the two garments can be identified. Matches may be made based on unique characteristics, such as wear marks and stitching patterns; however, most "new" garments are indistinguishable from all other new garments of the same type and size, rendering them of little evidentiary value. For criminal investigations that involve military personnel, however, the widespread use of standard-issue camouflage uniforms is a unique circumstance in which statistical garment association is feasible. In such a case, the garments are well specified, and the manufacturing process is standardized. This makes it possible to develop a statistical model that accounts for all significant sources of variation in the garment. The single most significant parameter in the statistical model developed is whether or not the two garments analyzed were manufactured with the same or similar markers. If the two garments can reasonably be assumed to come from independent markers, then the probability of an accidental match becomes negligible so long as three or more pieces are observed with reasonable precision in the surveillance image. Under the proposed model, an upper-bound on the incidence rate is calculated, which is compatible with current practices in evaluating statistical evidence in forensic science. The statistical model was developed for U.S. Army digital camouflage uniforms for forensic investigations by the U.S. Army Criminal Investigation Laboratory; however, the model can be used in any investigation that involves surveillance images of camouflage garments in which the camouflage pattern and garment manufacturing parameters are known. 2 tables, 12 figures, and 18 references