We present a high-statistics Monte Carlo determination of the exponent gamm
a for self-avoiding walks on a Manhattan lattice in two dimensions. A conse
rvative estimate is gamma greater than or similar to 1.3425 +/- 0.0003, in
agreement with the universal value 43/32 on regular lattices, but in confli
ct with predictions from conformal field theory and with a recent estimate
from exact enumerations. We find strong corrections to scaling that seem to
indicate the presence of a non-analytic exponent Delta < 1. If we assume D
elta = 11/16 We find gamma = 1.3436 +/- 0.0003, where the error is purely s
tatistical.