Nanonetworks consist of nano-sized communicating devices which are able to perform simple tasks at the nanoscale. The limited capabilities of individual nanomachines and the Terahertz (THz) band channel behavior lead to error-prone wireless links. In this paper, a cross-layer analysis of error-control strategies for nanonetworks in the THz band is presented. A mathematical framework is developed and used to analyze the tradeoffs between Bit Error Rate, Packet Error Rate, energy consumption and latency, for five different error-control strategies, namely, Automatic Repeat reQuest (ARQ), Forward Error Correction (FEC), two types of Error Prevention Codes (EPC) and a hybrid EPC. The cross-layer effects between the physical and the link layers as well as the impact of the nanomachine capabilities in both layers are taken into account. At the physical layer, nanomachines are considered to communicate by following a time-spread on-off keying modulation based on the transmission of femtosecond-long pulses. At the link layer, nanomachines are considered to access the channel in an uncoordinated fashion, by leveraging the possibility to interleave pulse-based transmissions from different nodes. Throughout the analysis, accurate path loss, noise and multi-user interference models, validated by means of electromagnetic simulation, are utilized. In addition, the energy consumption and latency introduced by a hardware implementation of each error control technique, as well as, the additional constraints imposed by the use of energy-harvesting mechanisms to power the nanomachines, are taken into account. The results show that, despite their simplicity, EPCs outperform traditional ARQ and FEC schemes, in terms of error correcting capabilities, which results in further energy savings and reduced latency.