Identifying the forces responsible for the origin and maintenance of sexuality remains one of the greatest unsolved problems in biology. The mutational deterministic hypothesis postulates that sex is an adaptation that allows deleterious mutations to be purged from the genome; it requires synergistic interactions, which means that two mutations would be more harmful together than expected from their separate effects. We generated 225 genotypes of Escherichia coli carrying one, two or three successive mutations and measured their fitness relative to an unmutated competitor. The relationship between mutation number and average fitness is nearly log-linear. We also constructed 27 recombinant genotypes having pairs of mutations whose separate and combined effects on fitness were determined. Several pairs exhibit significant interactions for fitness, but they are antagonistic as often as they are synergistic. These results do not support the mutational deterministic hypothesis for the evolution of sex.