Feedback between optical elements can be a major source of noise when trying to attain high sensitivity in infrared absorption experiments. We find that a conventional White‐cell optical arrangement introduces e´taloning fringes that modulate the peak‐to‐peak amplitude of our signals by 1 part in 16 666, a fractional change of 6×10−5. Although relatively small, this ‘‘noise’’ is systematic and adds coherently with averaging, obscuring interesting absorption features. An easily constructed multipass optical system suited for performing high‐resolution infrared spectroscopy in molecular beams is described. The design is based on a variation of the White cell and has been optimized for use with lead salt diode lasers. One of the key components in the improved design is the addition of an oscillating mirror for spoiling optical feedback generated by laser scatter and/or poor mode coupling of the laser to the multipass optics.