Discontinuation of benzodiazepine (BZD) treatment for insomnia can be a difficult task. Cognitive-behavior therapy (CBT) for insomnia, combined with a supervised medication taper, can facilitate withdrawal but there is limited evidence on long-term outcome after discontinuation. The objective of this study was to examine medication-free survival time and predictors of relapse (i.e., resumed BZD hypnotics) over a 2-year period in 47 older adults (mean age 62.1 years) with persistent insomnia and prolonged BZD use (average duration of 18.9 years), who had successfully discontinued BZD following CBT for insomnia, a supervised medication taper program, or a combined approach. The Kaplan-Meier product-limit method was used to estimate survival time, defined as time between end-of-treatment and relapse or end of follow-up. By the end of the 24-month follow-up, 42.6% of the samples had resumed BZD use. Participants in the Combined (33.3%) and Taper (30.8%) groups relapsed significantly less than their counterparts from the CBT group (69.2%). Survival rates at 3 months were 61.5% (CBT), 100% (Taper), and 80.9% (Combined). At 12 months, they were 38.5%, 83.3%, and 70.8%, respectively; and, at 24 months, they were 28.9%, 64.8% and 64.9%, respectively. Mean survival time was significantly longer for both the Taper (18.6 months, SE = 2.1) and Combined groups (12.6 months, SE = 1.4), relative to the CBT group (8.5 months, SE = 1.8). Significant predictors of relapse included treatment condition, end of treatment insomnia severity, and psychological distress. In conclusion, there is a substantial relapse rate following BZD discontinuation among prolonged users. CBT booster sessions might enhance compliance with CBT and prove useful in preventing relapse.