Differences and similarities between intrusive thoughts typical of obsessive–compulsive disorder, generalized anxiety disorder, and hypochondriasis are relevant for their differential diagnosis, formulation, and psychological treatment. Previous research in non‐clinical samples pointed out the relevance of some process variables, such as responsibility, guilt, or neutralization strategies. This research is aimed to investigate the differences and similarities between clinical obsessions, worries, and illness intrusions in some of these process variables. A second aim is to identify models based on these variables that could reliably differentiate between them. Three groups of patients with obsessive–compulsive disorder (n = 35; 60% women, mean age 38.57), generalized anxiety disorder (n = 36; 61.1% women, mean age 41.50), and hypochondriasis (n = 34; 70.6% women, mean age 31.59) were evaluated using the Cognitive Intrusions Questionnaire—Transdiagnostic Version (Romero‐Sanchiz, Nogueira‐Arjona, Godoy‐Ávila, Gavino‐Lázaro, & Freeston, ). The results showed that some appraisals (e.g., responsibility or egodystonicity), emotions (e.g., guilt or insecurity), neutralization strategies, and other variables (e.g., verbal content or trigger from body sensation) are relevant for the discrimination between obsessions, worries, and illness intrusions. The results also showed 3 stable models based on these variables for the discrimination between these thoughts. The implication of these results in the diagnosis, formulation, and psychological treatment of obsessive–compulsive disorder, generalized anxiety disorder, and hypochondriasis is discussed. Key Pratictioner Message The current study provides evidence about the discrimination between clinical obsessions, worries, and illness intrusions from a transdiagnostic perspective. The differentiation between them is crucial for the diagnosis, formulation, and psychological treatment of obsessive–compulsive disorder, generalized anxiety disorder, and hypochondriasis. Appraisals such as responsibility or egodystonicity, emotions such as guilt or insecurity, and other variables such as frequency, neutralization strategies or verbal form were found relevant for the discrimination between these types of thoughts. Three stable binary logistic models including these variables were developed for the discrimination between these thoughts.