Abstract For several years, the Institut de recherche Robert-Sauvé en santé et en sécurité du travail (IRSST) has been producing and publishing statistical indicators on occupational injuries (frequency, severity, frequency-severity, costs). They are usually produced every five years and are based on administrative data from the Commission des normes, de l’équité, de la santé et de la sécurité du travail (CNESST) covering, on average, a three-year period. The indicators are therefore based on data that do not reflect all the cumulative effects of occupational injuries. The assumption is that a three-year period yields enough data to be representative of all the data existing in closed compensation case files. It should also be noted that there are practical considerations associated with the use of a three-year data period, namely that there must not be too long a delay between the indicator reference years and today’s reality. This study examined, first of all, how using a longer data period would affect the indicators produced at the IRSST. Next, those same indicators were used to evaluate the stability of the classifications identifying the target groups and problems, in order to determine the soundness of using three-year-old data. Generally speaking, using nine years of data—three times longer than the usual period—often has considerable impacts that vary from one indicator to the next. When the data period was extended from three years to nine, the average compensation period per injury rose from 88.9 to 128.7 days—an increase of some 45% in six years. The frequency-severity rate rose from 3.0 to 4.3 days per FTE (full-time equivalent employee), which is also a 45% increase. Adding six years to the data resulted in an increase of about 44% in expenses for all cases. Overall costs (financial and human) did not increase in the same proportion, but they did show an increase of about 35%, from $4.7 billion to $6.3 billion (2015 dollars). The proportion of injuries requiring at least 180 days of compensation is one indicator that remained nearly unchanged by the extension, rising only from 10.5% to 10.7%. Despite the large changes in most of the indicators, extending the data period had little impact on the classification of injury descriptors or target groups (industries and occupations). There was very little change in rankings, based on the average cost per injury, or in the classification by injury descriptor (part of body injured, nature, type, cause). Some of these descriptors can be associated with target problems. The injuries most influenced by the extension were back injuries and psychological problems; these are the ones that advanced the most in the rankings, and for which the average cost per injury and total cost increased the most, relatively speaking. Whether according to average length of compensation, proportion of injuries with time lost and compensation of more than 180 days, proportion of injuries with permanent physical or psychological damage, or the frequency-severity rate of injuries with time lost, most of the target groups identified using a three-year data period remained the same when a nine-year period was used. Moreover, none of the target groups stood out significantly by achieving a higher or lower rank. This is true of both men and women. The study also shows which types of injuries had the highest increases in costs when the data period was extended by six years. Specifically, the cost increase (absolute and relative) was much greater for workers benefiting from rehabilitation and for cases involving a relapse, recurrence or aggravation, compared with cases where such factors are absent. To a lesser extent, larger cost increases were also observed for injuries with permanent physical or psychological damage than for those without, as well as for occupational illnesses compared with occupational injuries. To sum up, this study confirms that the gains arising from the use of a nine-year rather than three-year data period have no significant impact on the identification of target groups or on the designation of the most costly types of injuries, even though the indicator values may change noticeably. The use of a three-year data period therefore seems appropriate and in line with the IRSST’s needs. Nonetheless, it should be kept in mind that some problems or types of injuries are more underestimated than others: in particular, back injuries, psychological injuries, and cases involving rehabilitation or relapse.