A study was undertaken to compare nutrient composition changes when anhydrous ammonia, viable bacterial or non-viable bacterial preparations were applied to alfalfa hay baled at various moisture levels. Alfalfa forage was baled at three moisture levels: 15-20%, identified as low (L); 20-25%, identified as medium (M); 25-30%, identified as high (H). Hay baled at the M and H moisture levels was treated with one of two lactic acid-producing bacterial inoculants, Super-Sile (SS) or Sila-bac (SB), a non-viable bacterial preparation (Cul) applied at the time of baling or anhydrous ammonia (NH3) which was applied to forage when placed in a stack. Hay was stored in stacks for 60 days. The moisture level at baling did not (P>0.05) significantly influence field dry matter (DM) and nutrient losses, and did not alter the composition of hay at baling. Moisture level at baling did influence (P<0.05) storage DM and nutrient losses. Use of NH3 resulted in the highest (P<0.05) DM, crude protein (CP) and neutral detergent fibre retention values during storage relative to forage left untreated or treated with bacterial preparations. The application of NH3 did not improve (P>0.05) DM intake or DM, CP and acid detergent fibre digestibility of hay. Use of viable and non-viable lactic acid-producing bacterial preparations had variable results. Bacterial preparations did not improve (P>0.05) forage DM retention to the equivalent of NH3 for hay baled and stacked at the M and H levels. CP and soluble N content in hay treated with bacterial preparation were lower (P<0.05) than for hay treated with NH3 and similar to untreated hay baled at the L moisture level. DM and CP digestibilities were greater (P<0.05) for untreated hay when baled at L versus M moisture levels. Treatment of M hay with SB or Cul increased (P<0.05) digestibility values to the equivalent of L hay. SS treatment of M hay did not improve (P>0.05) DM and CP digestibility relative to untreated M hay. © 1990.