Dead-end and cross-flow organic solvent nanofiltrations (OSNs) using StarMem (TM) 122 (220 g mol(-1) nominal molecular weight cut-off (MWCO)) were conducted on a homologous series of trialkylamine bases, ranging from triethylamine (MW = 270 g mol(-1)) to tridodecylamine (TDDA; MW = 522 g mol(-1)). Dead-end OSN gave the expected rejections of 97-99% for bases larger than the MWCO, except TDDA, which unexpectedly had an average rejection of only 19%. MWCO therefore does not accurately represent the selectivity of StarMem (TM) 122. To investigate the reasons for this, cross-flow OSN of the same trialkylamine bases was conducted. This did not give a low TDDA rejection however, and the rejections of all the other bases were lower than in dead-end OSN. These results are most likely due to greater concentration polarisation and/or fouling in the dead-end OSNs. Furthermore, by reconciling these results with the currently understood membrane mass transport models, it is thought that it indicates that both pore flow and solution diffusion play a role in the trialkylamine transport across StarMem (TM) 122 membranes.