SummaryIn three experiments, ryegrass and lucerne were ensiled for 100 days in test-tube silos containing about 100 g of crop either without additive or after the addition of paraformaldehyde or formalin to provide 0·1–0.4% of the fresh crop weight as formaldehyde (HCHO). When applied as prills containing 82% HCHO, paraformaldehyde was as effective as formalin in restricting fermentation and preventing extensive protein breakdown at comparable HCHO application rates. The lower solubility of paraformaldehyde containing 98 % HCHO as powder or prills made these forms less effective than formalin in restricting fermentation. Paraformaldehydecontaining 98% HCHO was, however, effective even at the lowest rate of application, in preventing clostridial-type fermentations when these were present in silages made from untreated crops.This was in contrast to formalin which promoted such fermentation at low rates of application.