Based on urban microcell channel measurements, the bit error rate performance of a space-time block code with four transmit antennas using various dual-branch receive antenna configurations, is evaluated. It is demonstrated that with realistic handset antennas, it is possible to achieve performance very close to that of the theoretical uncorrelated Rayleigh case, although proximity of operator tissue (e.g. head) will result in several decibels performance degradation.Introduction: The performance of coded data transmission over multiple-input multiple-output (MIMO) channels depends on the spatial characteristics of antennas and the propagation environment. Even so, it is typically evaluated using idealised channel models, that are unable to capture the combined effect of real-world radio channels and handset antennas on wireless link performance. The three main factors degrading performance of real-world MIMO communications systems are spatial correlation, channel coefficient power imbalance, and SNR degradation owing to unfavourable antenna orientation. In particular, field patterns of realistic mobile handset antennas are not omnidirectional, and the received signal typically arrives from few dominating directions, or angular clusters. As a consequence, directivity and orientation of the antenna patterns play an important role in estimating system performance. In this Letter, we evaluate the effect of antennas on the bit error ratio (BER) performance of a quasiorthogonal space-time block code (QSTBC) for four transmit antennas in a practical urban microcell scenario.