In this article, we propose and study the performance of low-power impulse radio ultra-wideband correlation-based time-of-arrival (TOA) estimators for body area networks. In particular, we study and compare the signal-to-noise-ratio degradation of TOA correlatorbased estimators assuming optimal, real and complex suboptimal sinusoidal templates. Moreover, we study the improved Ziv-Zakai lower bound for complex suboptimaltemplate based TOA estimators. Then, we study and compare the performance of practical correlator-based TOA estimators using optimal, real and complex suboptimal sinusoidal templates via simulations in the IEEE 802.15.6a UWB on-body channel model as well as actual on-body measurements. The performance of real suboptimal template-based estimators approaches the performance of optimal TOA estimators based on simulations and actual measurements. We show that low-power is traded for a minimal performance loss.