Research is increasingly focusing on diagnostic errors and strategies to reduce them. The challenges of measuring diagnostic difficulties has hindered progress. This commentary outlines a conceptual approach to identifying "undesirable diagnostic events." The authors propose developing a list of clinical contexts and specific diseases prone to diagnostic error. Candidate conditions should be diagnosable in routine practice with a clear reference standard and defined diagnostic process. They also contend that measures should be constructed for relatively common conditions that are often misdiagnosed and for which delayed diagnosis could lead to harm, such as delayed cancer diagnosis. The authors propose designing and testing diagnosis measures based on this framework. A previous PSNet perspective by the senior author, Hardeep Singh, discussed momentum in the field of diagnostic error over the past several years.