Dermatoscopy is not accepted by all dermatologists as a useful diagnostic tool. We set out to test if training followed by regular clinical use of dermatoscopes improved diagnostic accuracy in melanoma diagnosis. Six dermatologists who had not previously used dermatoscopes were studied before and after a 10-month period of dermatoscope use with training and use of the Modified Pattern Analysis Diagnostic Algorithm. Diagnostic accuracy was assessed using test cards containing clinical and dermatoscopic photographs. The number of melanomas undiagnosed after training fell from 18 to 5 and the number potentially left unexcised fell from 18 to 3. The numbers of benign lesions potentially excised remained unchanged. The study shows that the use of dermatoscopes with training greatly increased the accuracy of diagnosis of melanoma by dermatologists. This practical study supports the use of dermatoscopy in pigmented-lesion diagnosis and demonstrates how dermatoscopy training could be incorporated into UK specialist training programmes.