Abstract Introduction Sleep Restriction Therapy (SRT) is one of the core components of CBT for Insomnia (CBT-I). However, there is little empirical evidence regarding how SRT implementation evolves over the course of therapy. Here, we propose a conceptual model for defining stages of SRT and assess the applicability of the model in a recently completed clinical trial. Methods Our proposed two-stage Build-Maintain model of SRT is: 1) The Build stage starts with SRT implementation. The primary goal is building homeostatic pressure until average weekly sleep efficiency (SE) is high enough to increase time-in-bed (e.g., 85% or 90%). 2) The Maintain stage starts with the first increase of time-in-bed after the target SE is reached. The goal is to maintain sleep pressure sufficiently high that one can maintain SE in the desired range. We assessed the model in 83 adults (age=48.3±14.8 years; 50F) undergoing CBT-I in a RCT. SRT was introduced in week 2. We explored the extent to which age, sex, psychiatric comorbidities, use of hypnotics, and baseline insomnia severity predicted progression through the SRT stages. Results Participants were in the Build stage 2.1±1.3 weeks (41.8% were in Build for only one week; 84% for ≤3 weeks). Higher baseline ISI (r=.238,p=.036) and lower baseline SE (r=-.365,p=.001) predicted spending more weeks in the Build stage before transitioning to the Maintain stage. Discussion The two-stage Build-Maintain model of SRT may inform: a) setting empirically-based expectations for client progress; b) examining adherence to CBT-I; c) evaluating shortened versions of CBT-I.