Long-term potentiation (LTP) is probably the most widely studied form of synaptic plasticity in the mammalian central nervous system. In the early descriptions, the term referred to a sustained increase in synaptic response following a brief high-frequency electrical tetanus. Apparently unique properties of the phenomenon triggered considerable excitement in the field: for many, LTP offered the promise of a potential substrate for learning and/or memory. In the more than 20 years since LTP was first discovered, investigators motivated by this promise have described a vast array of molecules and processes that may be involved in LTP induction and maintenance. And yet, the mechanisms by which LTP occurs have not been resolved. Instead, the compiled results have uncovered layer upon layer of intricacy, including multiple LTP forms and multiple molecular cascades involved in LTP expression. The generally stated thesis that LTP equates to learning and/or memory at a synaptic level has not faced a serious challenge despite the fact that workers in the field have not provided an unambiguous correlation of LTP with either. A number of investigators have now shifted their attention to a newer form of synaptic modification, long-term depression (LTP). Whatever studies of LTD reveal, it is clear that the fundamental questions about LTP remain unanswered: what is it really and what, if anything, is it used for? In this review, we summarize the data concerning putative LTP mechanisms and the evidence for LTP's role in learning and memory. We show that extant models are not sufficient to account for the various forms of LTP and that the experimental evidence does not justify the view that LTP equates to learning and memory. Instead, we suggest that LTP can be related to other forms of synaptic modification, e.g., LTD and kindling, in a neuroplasticity/pathology continuum of events. In particular, we suggest that neurotransmitter receptor regulation may be a key element leading to synaptic modification: in the adult nervous system, homeostatic receptor regulation normally compensates for alterations in synaptic input, while in the developing nervous system a form of 'homeodynamic' receptor regulation prevails. Our model proposes that homeodynamic receptor regulation leading to an LTP-like effect triggers, or acts in concert with, synaptogenesis to allow young neurons to modify response characteristics in response to altered input. In contrast, some forms of LTP in adult neurons may represent a 'failed' form of receptor regulation whose final outcome is neural death. The model suggests a series of experimentally verifiable hypotheses.