Virtually every prosthodontic rehabilitation involves making diagnostic casts for analysis and prosthesis fabrication. Frequently, these casts are produced using irreversible hydrocolloid (IH) impression materials in stock metal or plastic trays. However, it is unclear whether one technique produces a more accurate cast. The purpose of this randomized clinical trial was to compare the linear accuracy of diagnostic casts produced using IH with 1 of 3 different tray types: (1) perforated metal trays, (2) stock plastic trays, and (3) directed-flow stock plastic trays. All groups were compared to casts produced with custom trays and vinyl polysiloxane (VPS) impression materials, which were considered the control. Seven subjects participated in this trial. IH impressions were made in a random order using 1 of 3 tray types: stock plastic, perforated metal, or plastic directed flow. These were compared to VPS impressions using custom trays (control group). Each impression technique was repeated 3 times per subject, for a total of 84 observations. Impressions were disinfected and poured in a type IV stone. Linear accuracy of casts was measured using computer software analysis of scanned images of the casts at x30 magnification. Three linear measurements were made on each cast: second molar to second molar, right second molar to left first premolar, and left second molar to right first premolar. Measurements were compared among techniques using mixed-model analysis of variance to account for correlation among the multiple measurements made on each subject. Dunnett's adjustment for multiple comparisons with control was used (α=.05). For molar-to-molar and right second molar to the left first premolar measurements, there were no significant differences in linear dimensions between casts made from different trays. However, linear measurements from the left second molar to the right first premolar demonstrated significant differences for casts made with stock metal, directed-flow, and stock plastic trays compared to custom trays. In this group, casts produced by stock metal, directed-flow, and stock plastic trays differed from controls by 102, 68, and 71 um, respectively. Generally, casts made with plastic trays (stock plastic and directed flow) had values closer to those of custom trays than did casts made with metal trays. Impressions made with irreversible hydrocolloid produced casts that were significantly different in linear dimension than casts produced by custom trays and VPS impressions. These differences were not uniform, but varied by location on the cast. © 2010 The Editorial Council of the Journal of Prosthetic Dentistry.