A confocal epifluorescence detection scheme, optimized to give sub-picomolar detection within planar glass substrates etched to a 30 µm depth, is described. A ×40, 0.6 numerical aperture (N.A.) lens with a 3.7 mm working distance was used to create a focused laser spot about 12 µm in diameter, by under-filling the lens aperture to give an effective, measured N.A. of 0.22 for the laser beam. The sectioning power (optical axis field of view) of various pinholes and the corresponding detector probe volumes (overlap of the excitation and observation volumes) were: (pinhole diameter, sectioning power, probe volume): 100 µm, 18 µm, 0.1 pl; 200 µm, 20 µm, 0.4 pl; 400 µm, 26 µm, 1.7 pl; and 600 µm, 36 µm, 2.4 pl. A log–log plot of fluorescence intensity versus fluorescein concentration, measured in continuous-flow mode using the optimum 400 µm pinhole, showed a correlation coefficient of 0.996 and a slope of 0.85. In this mode, 300 fM fluorescein gave a signal of 34.6 ± 8.1 mV over background with an S/N of 6.1, representing the lowest measured fluorescein dye concentration reported on-chip. Capillary zone electrophoresis of 1 pM fluorescein resulted in a mean S/N of 5.8. The injection plug, estimated to be about 5470 molecules, corresponds to 570 detected molecules on average. The design and use of quick-fit, flangeless fittings for interfacing tubing, fused-silica capillaries or pressurized systems to microfluidic channels etched in planar glass chips is briefly presented.