Spontaneous lipolysis (SL) impairs the technological and sensory properties of milk and dairy products. Husbandry factors, and especially nutritional factors, are known to modulate SL levels, but the biochemical mechanisms involved remain unclear. We investigated effects of feeding levels and forage types on SL mechanisms. Thirty-two cows were divided into 4 groups according to their diets (corn silage or conserved grass). Two feeding levels were applied: "non-restricted", where cows were fed ad libitum, and "restricted", where cows were fed at 75% of their ad libitum dry matter intake. The trial was conducted in a split-plot design where "forage type" was the main factor and "feeding level" was the sub-factor. Cows were followed for 12.5 weeks: 4 pre-experimental weeks, then a 14-day transition period (to change the forage, and decrease feed intake), a first 3-week experiment period, then a 4-day transition period (to reverse the energy treatment) followed by a second 3-week experiment period. Milk samples were collected from morning and evening milkings at the end of each period to determine milk traits, fat globule size, protein and fatty acid profiles, lipoprotein lipase (LPL) activity, and SL. Data analysis using the SAS-software MIXED procedure found no statistically significant feeding-level x forage-type interactions on SL. Compared to non-restricted cows, feed-restricted cows tended to have higher SL levels in morning milks (0.60 vs 0.52 mEq/100 g fat) and even higher SL levels in evening milks (0.89 vs 0.58 mEq/100 g fat). Cows fed corn silage tended to have higher SL in morning milks (0.72 vs 0.40 mEq/100 g fat) but evening milks showed no forage-type effect. The low feeding level had no effect on milk fat content, fat globule size, LPL activity or protein profile. In both non-feed-restricted and feed-restricted cows, corn silage was associated with lower milk fat content (3.53 vs 3.91%), smaller globule size (d(4,3) : 3.40 vs 3.71 mu m), lower percent proteose peptone 5 and 8 s, but had no effect on LPL activity. Data analysis using the SAS-software REG procedure found that proteose peptone 5 (logarithm-converted) was strongly negatively correlated with SL, suggesting a potential inhibitor effect of proteose peptone 5. Certain milk components are therefore modulators of SL. In morning and evening milks of SL-sensitive cows (defined as SL level up to 0.60 mEq/100 g fat), SL was negatively correlated with C18:0 and fat content and positively correlated with cis9 C18:1/C18:0, suggesting a link between tissue mobilization, mammary metabolism and SL. To sum up, feeding restriction is associated with higher SL, probably due to higher tissue lipid mobilization. Further investigation is needed to determine the impact of corn silage on SL. Variations in proteose peptone 5 levels according to forage type could explain SL levels. These findings could serve for advisory practice and prevention education to farmers.. At a biochemical level, the findings could be used to create tools and tests to detect and monitor lipolysis cow-by-cow.