The mutation, prmE37, located at -14 in the PRM promoter of bacteriophage lambda, reduces PRM function dramatically both in vitro and in vivo. In a search for second-site revertants of prmE37, we isolated a double mutant that exhibits a partially restored Prm+ phenotype. The second-site mutation (at -31) is identical to the mutation prmup-1. The activity of the doubly mutant (pseudo-revertant) promoter, prmE37prmup-1, was investigated in vivo using a PRM-lacZ fusion phage and found to be intermediate between that of prmE37 and wild-type PRM. However, the relative strength of the prmE37prmup-1 promoter was greater than expected following superinfection of a lambda lysogen. Since nalidixic acid was found to preferentially inhibit transcription from the doubly mutant promoter under these conditions, we suggest that DNA supercoiling favors activation of this promoter by repressor. In runoff transcription assays in the absence of repressor, the activity of wild-type PRM and the doubly mutant promoter were the same. However, while addition of repressor significantly stimulated wild-type PRM, it had little or no effect on the activity of the doubly mutant promoter. Values of KB, the equilibrium constant for formation of closed complexes, and kf, the rate constant for isomerization of closed to open complexes, were determined in abortive initiation assays, and the product of kfKB was used as a measure of promoter strength. The results of these assays are in agreement with those obtained in runoff transcription assays. In the absence of repressor, values of kfKB for the doubly mutant promoter and wild-type PRM are the same; however, tau obs, the time required for open complex formation, is significantly greater for the double mutant than for wild-type PRM at all RNA polymerase concentrations used for the abortive initiation analysis. In the presence of repressor, the doubly mutant promoter is stronger than the prmE37 promoter, but much weaker than wild-type PRM. This is due to the fact that kf for the doubly mutant promoter is increased 2.5-fold by repressor, but KB is reduced to the same extent. These two effects counteract each other, so that repressor has no net effect on the strength of the prmE37prmup-1 promoter in vitro. In contrast, repressor increases kf for wild-type PRM eightfold and increases overall promoter strength (KBkf) nearly fivefold. In the presence of repressor, the effects of the two mutations, prmE37 and prmup-1, on kf are independent. This observation is discussed in relation to revised models for open complex formation.