Specific and processive antitermination by bacteriophage lambda N protein in vivo and in vitro requires the participation of a large number of Escherichia coli proteins (Nus factors), as well as an RNA hairpin (boxB) within the nut site of the nascent transcript. In this study we show that efficient, though nonprocessive, antitermination can be induced by large concentrations of N alone, even in the absence of a nut site. By adding back individual components of the system, we also show that N with nut+ nascent RNA is much more effective in antitermination than is N alone. This effect is abolished if N is competed away from the nut+ RNA by adding, in trans, an excess of boxB RNA. The addition of NusA makes antitermination by the N-nut+ complex yet more effective. This NusA-dependent increase in antitermination is lost when delta nut transcripts are used. These results suggest the formation of a specific boxB RNA-N-NusA complex within the transcription complex. By assuming an equilibrium model, we estimate a binding constant of 5 x 10(6) M-1 for the interaction of N alone with the transcription complex. This value can be used to estimate a characteristic dissociation time of N from the complex that is comparable to the dwell time of the complex at an average template position, thus explaining the nonprocessivity of the antitermination effect induced by N alone. On this basis, the effective dissociation rate of N should be approximately 1000-fold slower from the minimally processive (100-600 bp) N-NusA-nut+ transcription complex and approximately 10(5)-fold slower from the maximally processive (thousands of base pairs) complex containing all of the components of the in vivo N-dependent antitermination system.