In one-step growth experiments we found that bacteriophage Mu grew less efficiently in nonreplicating dnaA mutants than in dnaA+ strains of Escherichia coli. Phage development in dnaA hosts was characterized by latent periods that were 15 to 30 min longer and an average burst size that was reduced by 1.5- to 4-fold. The differences in phage Mu development in dnaA and dnaA+ strains were most pronounced in cells infected at a low multiplicity and became less pronounced in cells infected at a high multiplicity. Many of these differences could be eliminated by allowing the arrested dnaA cells to restart chromosome replication just before infection. In continuous labeling experiments we found that infected dnaA strains incorporated 5 to 40 times more [methyl-3H]thymidine than did uninfected cells, depending on the multiplicity of infection. DNA-DNA hybridization assays showed that greater than 90% of this label was contained in phage Mu DNA sequences and that only small amounts of the label appeared in E. coli sequences. In contrast, substantial amounts of label were incorporated into both host and viral DNA sequences in infected dnaA+ cells. Although our results indicated that phage Mu development is not absolutely dependent on concurrent host chromosomal DNA replication, they did strongly suggest that host replication is necessary for optimal growth of this phage.