Yes, in my experience the best procedure is this:
1. Record the samples in 24-bit or 32-bit fp-format
2. Normalize all samples to 0.3 dB
3. Convert all samples to 16-bit
4. Cut all samples accurately
5. Some sample creators convert the samples to FLAC or OGG.
The eveness and consistency in volume is achieved
best in the sample-playback definition inside the
sampler - that means in the formats themselves:
NKI, EXS, FXP or sfz.