Yes, a perfect recording of the amplitude of a function at a frequency of at least 2B contains enough information to perfectly reconstruct that function. Reaching this frequency is not that hard, but getting a perfect amplitude measurement is. This is where "noise, distortion and dynamic range" come in, they act as disturbances to the amplitude.