Quantitative assessment of the spontaneous or induced genomic mutation rate, a fundamental evolutionary parameter, usually requires the use of well-characterized mutant selection systems. Although there is a great number of genetic selection schemes available in Escherichia coli, the selection of D-cycloserine resistant mutants is shown here to be particularly useful to yield a general view of mutation rates and spectra. The combination of a well-defined experimental protocol with the Ma-Sandri-Sarkar maximum likelihood method of fluctuation analysis results in reproducible data, adequate for statistical comparisons. The straightforward procedure is based on a simple phenotype-genotype relationship, and detects mutations in the single-copy, chromosomal cycA gene, involved in the uptake of D-cycloserine. In contrast to the widely used rifampicin resistance assay, the procedure selects mutations which are neutral in respect of cell growth. No specific genetic background is needed, and practically the entire mutation spectrum (base substitutions, frameshifts, deletions, insertions) can simultaneously be measured. A systematic analysis of cycA mutations revealed a spontaneous mutation rate of 6.54 x 10(-8) in E. coli K-12 MG1655. The mutation spectrum was dominated by point mutations (base substitutions, frameshifts), spread over the entire gene. IS insertions, caused by IS1, IS2, IS3, IS4, IS5 and IS150, represented 24% of the mutations.