This article reviews empirical findings on two key premises of the appraisal model of obsessive-compulsive disorder (OCD): (a) non-clinical populations experience intrusive thoughts (ITs) that are similar in form and in content to obsessions; and (b) ITs develop into obsessions because they are appraised according to dysfunctional beliefs. There is support for the universality of ITs. However, the samples used are not representative of the general population. IT measures do not relate systematically or exclusively to OCD symptom measures, and are not specific enough to exclude other types of intrusive thoughts such as negative automatic thoughts or worries, nor are they representative of all types of obsessions. When general distress is controlled, there is so far no evidence that participants with OCD endorse obsessive belief domains more strongly than anxious participants, and inconclusive evidence that OCD and non-clinical samples differ on the belief domains. Some OCD symptom subtypes are associated with belief domains. Currently, there is no coherent model to offer strong predictions about the specificity of the empirically derived belief domains in OCD symptom subtypes. Cognitive therapy based on the appraisal model is an effective treatment for OCD, although it does not add to the treatment efficacy of behaviour therapy. It is unclear how appraisals turn ITs into obsessions. Implications for future research are discussed.