The degree of severity of an injury of the knee joint is estimated mainly on the basis of the history given by the patient, stability testing and hemarthrosis. Once these are known the surgeon usually decides whether an arthroscopy should be done or not. In a group of 365 patients who had undergone arthroscopy after acute knee injury, we made a retrospective check of the indications. The purpose of the study was to evaluate the sensitivity of hemarthrosis as an indicator of severe injury of the knee joint necessitating operative treatment. For stability testing we used the varus and valgus stress test, the Lachman test, the Anterior drawer test, the pivot shift test and the anterior drawer test with medial and lateral rotation. The Lachman test was repeated under general anesthesia in the majority of patients just before the arthroscopy. The results were evaluated retrospectively with the aid of electronic data processing. In nearly 70% of cases the arthroscopy was indicated because of positive signs of instability together with a typical history provided by the patient or hemarthrosis. In another 27%, the hemarthrosis was the sole reason for the surgeon's decision to perform an arthroscopic investigation of the injured knee joint. In 80.5% of all cases the clinical diagnosis of lesion of the anterior cruciate ligament (ACL) or combined injury to ligaments and menisci was confirmed by the arthroscopy. In addition to this group, in another 10.8% we found other severe lesions necessitating by operation.(ABSTRACT TRUNCATED AT 250 WORDS)