The practical limitation on sensitivity in broad‐line NMR is frequently spurious effects on the baseline rather than the masking of the signal by thermal noise. I discuss some of the causes of spurious baseline effects and describe a simple circuit used to reduce them by a factor of 300.