The effects of a number of experimental parameters on the efficiency of ion transport by viscous gas flow through narrow capillaries have been studied. Both electrospray and corona ion sources were used. The experimental data are consistent with ions loss to the walls of the capillary, which initially is caused mainly by space-charge expansion, but later is caused by diffusion. These processes can result in severe discrimination against low mass ions. The extent of ion loss may be calculated by using a simple model for radial diffusional loss in long cylinders, with an exponential decay of the ion density along the transport capillary. However, such a simple model underestimates ion loss by ignoring the effects of space-charge, turbulent flow, and rapid decay of higher radial diffusion modes (enhanced loss of ions that enter the capillary close to the wall). In contrast, Monte Carlo simulations showed that the effect of the parabolic velocity profile, under laminar flow conditions, is to increase the trasnmitted ion current, sometimes by several orders of magnitude, relative to the predictions of the simple diffusion model. After considering all these factors, the transmitted current from a corona was well reproduced by using mobility values for ions formed in such discharges. However, the measured transmitted current from an electrospray source was much too high. To explain this, it was necessary to assume that about 2% of the electrospray current is carried by aerosol particles with radii in the 10-25-angstrom range. Finally, it is argued that in glass capillaries wall charging may explain why the transmitted ion current is observed to be very similar to that in metal capillaries.