Traditional results in information and communication theory assume that receivers perform without error.  Permanent and transient errors, however, provide fundamental limits to computation systems just as they do to communication systems.  In this talk, I will describe two kinds of receiver errors and determine limits on communication performance under each.  One type of error is catastrophic failure of the receiver at a random time, which may be caused by device failure, energy exhaustion, or permanent jamming.  I show how to maximize the number of bits communicated at a fixed error probability.  Another type of error results from transient faults in the receiver’s decoding hardware, which may be caused by on-chip interference or low-power operation.  I quantify the robustness of standard iterative decoders for low-density parity-check codes to transient decoding noise.  In closing, I describe the problem of transmitting energy along with information so as to reduce receiver unreliability.