The applicability of COST231, WINNER+, ITU-R, and 3GPP urban microcell outdoor-to-indoor path loss prediction models is empirically tested in 100 m radius urban femtocells. The analysis is based on more than 500 link measurements. Considerations not specified in the models such as the inclusion of antenna gains and the effect of base station location on the street are evaluated. We find that for our scenarios, discrepancies among the models can reach up to 10 dB at the median. Using the WINNER+ model path loss with free-space antenna gains yields the best match to our measurements.