Rarely are the occupational fortunes and aspirations of a sizeable, historically significant, but non-elite group, available to the historian. Fortunately information collected during the 1937 Youth Employment Survey in Victoria still survives and provides one such rich source of information. The intention behind the survey was to identify 'youths' between 18 and 25 who had missed opportunities for 'normal' employment or who had been educationally disadvantaged during the Depression years 1929-1934. The ultimate aim was to provide them with the opportunity for technical training. According to the organizers it was 'not necessary' for a 'lad to be unemployed to be considered in any proposed scheme', those employed in 'dead-end jobs' were also considered eligible (Giles, 1937, p.1). Although the survey was organized through a special government committee the government of the day had not, at the time, made any financial commitment to the training of the youths who responded. Furthermore, there was to be very little room for choice of training in the scheme that was set up a year later. The single-sided survey form simply sought the applicant's, age, address, educational qualifications, employment history and preference for occupational training. The main historical sources of data that have been located comprise four, out of a possible six, reports on the survey and subsequent training program, and a progress report on the first 1000 applicants given as evidence at the Conference of Commonwealth and State Ministers on Youth Employment, held in Melbourne on 5 February 1937. The original completed forms can not be located which means that it is necessary to rely on statistical summaries. Even so the existing information provides a unique profile of a self-selected population of young males actively seeking new opportunities, The problems associated with secondary statistical sources in history are many. They include inconsistent classification, lack of rationale for classification, missing data and errors in transcription and calculation. However, the complete absence of the latter and the relatively easy comparability of data reported at four different dates and at different points in the data analysis, inspire confidence in this particular data source. So too does a knowledge of the fastidiousness, more than average statistical skills and expertise in occupational analysis of the person who tabulated the data and prepared the reports (Holbrook,1990). Nonetheless the limitations of the raw data also have to be borne in mind, including the possibility that the respondents embellished the facts or misunderstood the survey questions. The simplicity of the survey tends to dispel most concerns about the latter, but given that youths may have had something to gain from making their cases seem worthy of special consideration, the reliability of the data must be further queried. One mitigating factor is the size of the response group. Over six thousand youths responded within the set period and a further two thousand at a later period. To conclude, the simplicity of the survey, the large numbers, and the consistency and accuracy of classification and calculation when taken together tend to suggest that generalizations can be made from the secondary sources with a reasonable amount of confidence.