Very little is known about the distribution of mutational effects on o
rganismal fitness, despite the fundamental importance of this informat
ion for the study of evolution. This lack of information reflects the
fact that it is generally difficult to quantify the dynamic effects of
mutation and natural selection using only static distributions of all
ele frequencies. In this study, we took a direct approach to measuring
the effects of mutations on fitness. We used transposon-mutagenesis t
o create 226 mutant clones of Escherichia coli. Each mutant clone carr
ied a single random insertion of a derivative of Tn10. All 226 mutants
were independently derived from the same progenitor clone, which was
obtained from a population that had evolved in a constant laboratory e
nvironment for 10,000 generations. We then performed competition exper
iments to measure the effect of each mutation on fitness relative to a
common competitor. At least 80% of the mutations had a significant ne
gative effect on fitness, whereas none of the mutations had a signific
ant positive effect. The mutations reduced fitness by about 3%, on ave
rage, but the distribution of fitness effects was highly skewed and ha
d a long, flat tail. A compound distribution, which includes both gamm
a and uniform components, provided an excellent fit to the observed fi
tness values.