This paper is aimed at describing and characterizing occupational injuries which do not receive compensation (minor injuries), and at evaluating the differences between compensated and non compensated injuries with reference to type and site of the lesion. Data were collected in a car manufacturing plant and covered the period 1984-1991. Information regarding all kinds of occupational accidents were extracted, following a standardized procedure, from different paper forms compiled by health personnel within the factory (physicians and nurses). Due to the large amount of information registered, a sampling scheme was adopted which considered the same three months every two years, previously selected to cover the entire observation period (1984-85, 1987-88, 1990-91). In addition, to evaluate che adequacy of the sampling procedures for the compensated injuries only, the resulting sampled distributions of the relevant variables were compared with the totality of the injuries. 7,299 events occurred in the sampled months, which corresponded to 652.59 occupational accidents per million hours worked: 8.97% of them caused absence from work lasting more than three days (compensated cases), 3.59% caused a less than three workday loss (LTTWL cases), and the great majority (87.44%) did not cause any absence from work (first aid cases). The frequency index of the total number of accidents did not change over time, but the three categories of injuries did change: for example, first aid cases had their maximum frequency during the middle period (with a peak of 93.21% in March 1987) and their minimum during the last period (73.28% in June 1991). An opposite pattern occurred for accidents which caused absence from work. Type and site of the lesions differed among the three categories of occupational accidents. For example, cuts/lacerations were observed in 48% of first aid cases, 30% of compensated cases, and 15% of LTTWL cases. On the contrary, injuries due to extraneous bodies and contusions which were the most frequent category for LTTWL cases with, respectively, 28% and 25% of the cases, represented only 10% and 13% of first aid cases. With respect to the site of the lesion, hands had the highest frequency index both for first aid (56%) and compensated (39%) cases, and eyes were most affected (36%) among LTTWL cases. With respect to compensated injuries, a notable correspondence of the distributions of the relevant variables between the sample and the whole population was evident, also when considering separate individual subperiods.(ABSTRACT TRUNCATED AT 400 WORDS)