Generalized transduction has for about 30 years been a major tool in the genetic manipulation of bacterial chromosomes. However, throughout that time little progress has been made in understanding how generalized transducing particles are produced. The experiments presented in this paper use phage lambda to assess some of the factors that affect that process. The results of those experiments indicate: the production of generalized transducing particles by bacteriophage lambda is inhibited by the phage lambda exonuclease (Exo). Also inhibited by lambda Exo is the production of lambda docR particles, a class of particles whose packaging is initiated in bacterial DNA and terminated at the normal phage packaging site, cos. In contrast, the production of lambda docL particles, a class of particles whose packaging is initiated at cos and terminated in bacterial DNA, is unaffected by lambda Exo; lambda-generalized transducing particles are not detected in induced lysis-defective (S-) lambda lysogens until about 60-90 min after prophage induction. Since wild-type lambda would normally lyse cells by 60 min, the production of lambda-generalized transducing particles depends on the phage being lysis-defective; if transducing lysates are prepared by phage infection then the frequency of generalized transduction for different bacterial markers varies over a 10-20-fold range. In contrast, if transducing lysates are prepared by the induction of a lambda lysogen containing an excision-defective prophage, then the variation in transduction frequency is much greater, and markers adjacent to, and on both sides of, the prophage are transduced with much higher frequencies than are other markers; if the prophage is replication-defective then the increased transduction of prophage-proximal markers is eliminated; measurements of total DNA in induced lysogens indicate that part of the increase in transduction frequency following prophage induction can be accounted for by an increase in the amount of prophage-proximal bacterial DNA in the cell. Measurements of DNA in transducing particles indicate that the rest of the increase is probably due to the preferential packaging of the prophage-proximal bacterial DNA. These results are most easily interpreted in terms of a model for the initiation of bacterial DNA packaging by lambda, in which the proteins involved (Ter) do not recognize any particular sequence in bacterial DNA but rather recognize some feature of the DNA tht is sensitive to lambda exonuclease, such as a nick or a double-stranded cut.(ABSTRACT TRUNCATED AT 400 WORDS)