Objective: To answer the question, why is it necessary lo give a frame of clinical practice based in a quantitative medical language. It is done through the evaluation of agreenment between the diagnostic of multiple myeloma (MM) based in clinical detection of the disease pattern and in simplified criteria accepted as diagnostic confirmation. Design: Case series study, descriptive and of agreenment. Place and time of study: Hematology Unit of the Faculty of Medicine, Universidad Nacional de Colombia and San Juan de Dios Hospital, in Santafe de Bogota since January, 1994 until December, 1999. Material and methods: Definitive diagnostic of MM was unknown at the moment it was made the clinical prediction. Following the initial clinical evaluation of each patient, the hematologist wrote his probability, in numbers, with which he believed it will be confirmed his diagnostic impressions with subsequent laboratory study, which included the simplified criteria to diagnose MM. To measure the predictive accuracy it was used the average predictive accuracy coeflicient (Q) given in percentage. Results: Patients included in the study were 16. The average predictive coefficient of initial clinical evaluation in patients with MM was 46%. Twelve patients were males and four females. The median of age for the entire group was 60.5 years, with limits of the interquartile range of 50 years and 65 years. The clinical manifestations with the greatest frecuency at consultation, each one informed by five patients, were lumbar pain, general musculoskeletal pain and weight loss. The frecuency of nosologic entities included in the differential diagnose, besides MM, were mestastasis of carcinoma in seven patients, and lymphoma and bone tuberculosis, each one in three patients. Discussion: The physician skill to perform diagnostic predictions of MM was median. MM diagnosis always requires to be falsified a posteriori by routinary laboratory. The best way to quantify the medical uncertainty is to employ the language of probability.