Anisotropy of dose distributions around 103Pd, 125I, and 192Ir sources for interstitial brachytherapy was examined. Dose rates around 125I models 6702 and 6711 and 192Ir sources were measured using lithium fluoride thermoluminescence dosimeters (LiF TLDs) in a water-equivalent, solid phantom. From these measured data for 125I and 192Ir and the previously published measured data for 103Pd, isodose rate contours were determined using a bivariate interpolation and smooth surface fitting algorithm. The anisotropy functions, F(r,theta), as defined by the Interstitial Collaborative Working Group (ICWG) for each source, were determined. Also, 4 pi-averaged anisotropy factors, phi an(r), for use in point source approximation, have been calculated at radial distances varying from 1-10 cm for 103Pd, 125I, and 192Ir sources. The anisotropy factors had average values of 0.90, 0.93, 0.95, and 0.98 for 103Pd, 125I model 6711, 125I model 6702, and 192Ir, respectively. The anisotropy factors determined from dose measurements in phantom are observed to be closer to unity than from those determined previously from in-air measurements. This can be attributed to the smoothing of two-dimensional dose distributions due to the presence of more scattered photons in the phantom measurements compared to in-air measurements. Because in-phantom measurements simulate more closely the brachytherapy patient, data from these experiments are recommended for a more accurate determination of dose distributions around clinical brachytherapy implants. In this work, we present a complete set of source data for two-dimensional dosimetry following the ICWG formalism.