Prompted by indications from QSO lensing that there may be more mass associated with galaxy groups than expected, we have made new dynamical infall estimates of the masses associated with 2PIGG groups and clusters. We have analysed the redshift distortions in the cluster-galaxy cross-correlation function as a function of cluster membership, cross-correlating z<0.12 2PIGG clusters and groups with the full 2dF galaxy catalogue. We have made estimates of the dynamical infall parameter beta and new estimates of the group velocity dispersions. We first find that the amplitude of the full 3-D redshift space cross-correlation function, xi_{cg}, rises monotonically with group membership. We use a simple linear-theory infall model to fit xi(sigma, pi) in the range 5<s<40h^{-1}Mpc. We find that the beta versus membership relation for the data shows a minimum at intermediate group membership n~20 or L~2x10^11h^-2Lsun, implying that the bias and hence M/L ratios rise by a significant factor (~5x) both for small groups and rich clusters. However, the mocks show a systematic shift between the location of the beta minimum and the M/L minimum at L~10^10h^-2Lsun given by direct calculation using the known DM distribution. Our overall conclusion is that bias estimates from dynamical infall appear to support the minimum in star-formation efficiency at intermediate halo masses. Nevertheless, there may still be significant systematic problems arising from measuring beta~1/b using large-scale infall rather than M/L using small-scale velocity dispersions.