The P1 plasmid addiction operon (a classic toxin-antitoxin system) encodes Phd, an unstable 73-amino-acid repressor-antitoxin protein, and Doc, a stable toxin. It was previously shown by deletion analysis that the N terminus of Phd was required for repressor activity and that the C terminus was required for antitoxin activity. Since only a quarter of the protein or less was required for both activities, it was hypothesized that Phd might have a modular organization. To further test the modular hypothesis, we constructed and characterized a set of 30 point mutations in the third and fourth quarters of Phd. Four mutations (PhdA36H, V37A, I38A, and F44A) had major defects in repressor activity. Five mutations (PhdD53A, D53R, E55A, F56A, and F60A) had major defects in antitoxin activity. As predicted by the modular hypothesis, point mutations affecting each activity belonged to disjoint, rather than overlapping, sets and were separated rather than interspersed within the linear sequence. A final deletion experiment demonstrated that the C-terminal 24 amino acid residues of Phd (preceded by a methionine) retained full antitoxin activity.