Criteria for the diagnosis of myocardial infarction vary not only from one cardiac intensive care unit (CICU) to another, but also from one study to another. Even the appropriate juncture for ECG, the type of biochemical markers used and blood sampling times vary. Thus, epidemiological studies comparing results over time or between various regions or hospitals tend to be misleading. Reported results are difficult to interpret and to apply to one's own CICU. In order to survey myocardial infarction diagnosis in Sweden and planned future changes, in February-March 1997 a questionnaire was sent to all 82 CICUs in the country. Of the 74 (90%) responders, 72% (53/74) reported formalized printed criteria for myocardial diagnosis to be available at the unit. Eight different biochemical markers of myocardial injury were in use; CK-MB (creatine kinase and its cardio-specific isoenzyme) was the most common, being used at 64% (47/74) of the units; CK and CK-B were used at 32%, and troponin T or I at 53%. Myoglobulin has not been very widely used. If planned changes are carried out, 86% of the units will soon be using CK-MB, and 79% troponin T och I. Cut-off levels of biochemical markers of myocardial infarction varied. Of the 47 units where CK-MB was used, the cut-off level was 10 micrograms/L at 10 (28%) of the units, 15 micrograms/L at 31 (66%) units, and a higher level in a smaller group of units. Cut-off levels for CK-B manifested a similar lack of uniformity. The greatest difference was manifested by troponin levels; of the 28 units using quantitative tests and that cited their cut-off levels, three (11%) used 0.10 microgram/L, 12 (43%) used 0.20 microgram/L, and the remaining 13 (46%) used 0.50 microgram/L. The use of ischaemia monitoring in conjunction with diagnosis and prognosis of myocardial infarction has increased, 72% of the units reporting that they used some form of monitoring, and a further 13% that they planned to introduce it in the near future. Thus, the questionnaire study showed marked differences in myocardial infarction diagnosis to exist in Sweden, although a manifest trend toward increasing uniformity was also seen, and the outlook for the standardisation of diagnostic criteria is good.