In a prospective controlled trial the effects of a 5-day in-patient treatment and teaching programme for Type 1 (insulin-dependent) diabetes mellitus on metabolic control and health care costs were studied in Moscow. Two different intervention programmes were compared, one based upon urine glucose self-monitoring (UGSM, n = 61) and one using blood glucose self-monitoring (BGSM, n = 60). Follow-up was 2 years. A control group (n = 60) continued the standard treatment of the Moscow diabetes centre and was followed-up for 1 year. Costs and benefits with respect to hospitalizations and lost productivity (according to average wage) were measured in November 1992 rubles (Rb.), with respect to imported drugs and test strips in 1992 German marks (DM). In the intervention groups there were significant decreases of HbA1 values [UGSM: 12.5% before, 9.4% after 1 year, 9.2% after 2 years (p < 0.0001); BGSM: 12.6% before, 9.3% after 1 year, 9.2% after 2 years (p < 0.0001) compared to no change in the control group (12.2% before, 12.3% after 1 year)], and of the frequency of ketoacidosis. The frequency of severe hypoglycaemia was comparable between the UGSM (10 cases during 2 years), BGSM (10 cases during 2 years), and the control group (8 cases during 1 year).(ABSTRACT TRUNCATED AT 250 WORDS)