The Drosophila salivary gland secretion protein gene, Sgs3, has a consensus TATA sequence and gives rise to abundant stage and tissue-specific transcripts. Two TATA point mutations (TAAA and TAGA) reduce transcript levels approximately 50-fold when assayed in transgenic flies. This effect is reflected in vitro, in DNase I footprint and gel retardation assays where we observed TATA-probe-specific complexes that are not seen with TAAA, TAGA or non-specific probes. The binding patterns observed when using nuclear extracts from 0-2- and 0-20-h embryos (Sgs3 inactive) differ from those seen with extracts from third instar salivary glands (Sgs3 active). There are also differences in in vitro binding when using an hsp70 TATA fragment, previously shown to substitute in vivo for the Sgs3 TATA sequence, as probe. Together these observations suggest the possibility that more than one TATA box factor may be present in these extracts. We conclude that a wild-type TATA motif is crucial for the binding of a TATA box factor and all subsequent interactions with other factors bound to the proximal and distal regulatory sequences that are necessary for the normal expression of Sgs3.