Traditional assumptions hold that being sent to court is a harsher sanction than being diverted from court, that screening decisions necessarily involve judgments about the individual juvenile offender, and that official records are useful data sources for studying the processing of cases. A study of probation officers' discretionary screening decisions at two family court intake units of a large suburban jurisdiction of New York State involved extensive interviews with intake workers and analysis of 500 completed juvenile case records. Apparent from this investigation is the fact that at least some of the time, intake workers regard court appearances as the more lenient screening option, base their screening decisions on considerations not directly related to the individual offender, and develop records that deliberately obscure their actual case handling. It was found that cases are being screened without interviewing the juvenile involved and that children sometimes enter the intake unit with one charged offense and are sent to court with several or with an altered offense description. Moreover, actual outcomes of screening decisions and adjustment efforts sometimes diverge considerably from the recorded outcomes. These bizarre inconsistencies result from the fact that intake offcers' discretional decisions reflect their own implicit theories of delinquency causation and control and their perceptions of the juvenile justice system's bureaucratic realities. Their working assumptions are that court referrals often result in more lenient and less effective treatment than out-of-court arrangements, that screening decisions should be based on the juvenile's present social situation rather than on judgments about the offender's personality and the specifics of the charge, and that the purpose of the official record is to facilitate positive intervention rather than to support facts. These assumptions must be taken into account in order to prevent misleading and confusing research results based on traditional theoretical assumptions. Footnotes and 38 references are included.