The Obsessive Compulsive Inventory (OCI) is often used as a screening instrument for symptoms of obsessive-compulsive disorder (OCD) and as an outcome measure for treatment. Three versions of the OCI are available: the original 42-item version, the revised 18-item version (OCI-R) and a shorter version that focuses on the highest subscale (OCI-R Main). Our aim was to determine sensitivity to change and evaluate cut-off scores for caseness in each version of the OCI using the same dataset. Method: We compared the effect size and the number of patients who achieved reliable and clinically significant change after cognitive behavior therapy in two samples of out-patients with OCD. One sample (n = 63) had OCD of minor to moderate severity and a second sample (n = 73) had severe, treatment refractory OCD. Results: In both samples, the OCI and OCI-R had very similar treatment effect sizes and to a lesser extent in the percentage who achieved reliable improvement and clinically significant change. The OCI-R Main was more sensitive to change than the OCI or OCI-R in both samples. All versions of the OCI were less sensitive to change compared with the Yale-Brown Obsessive Compulsive Scale (Y-BOCS). Discussion: The OCI-R is a valid self-report outcome measure for measuring change and is less burdensome for patients to complete than the OCI. Questions remain about whether the OCI or OCI-R is sufficiently sensitive to change for a service evaluation. We would recommend a slightly higher cut-off score of ≥17 on the OCI-R for the definition of caseness.