We present analytical and numerical results of a random walk on the family
of small-world graphs. The average access time shows a crossover from regul
ar to random behavior with increasing distance from the starting point of t
he random walk. We introduce an independent step approximation, which enabl
es us to obtain analytic results for the average access time. We observe a
scaling relation for the average access time in the degree of the nodes. Th
e behavior of the average access time as a function of p shows striking sim
ilarity with that of the characteristic length of the graph. This observati
on may have important applications in routing and switching in networks wit
h a large number of nodes.