In a storage experiment with dairy cow manure, the effects of dietary protein content and manure type on ammonia, nitrous oxide and methane volatilization as well as overall nitrogen (N) loss from manure were investigated. Early-lactating cows received rations with 175, 150 and 125 g crude protein/kg dry matter. Each ration was tested in four manure storage systems reflecting typical farm conditions. These either contained total excreta with high amounts of straw (deep litter manure) or no straw (slurry) or, proportionately, 0·9 of urine and 0·1 of faeces (urine-rich slurry) complemented by the residuals with a low amount of straw (farmyard manure). Manure samples were stored for 7 weeks under controlled conditions and trace gas emission was repeatedly measured. Reduction of N intake decreased daily N excretion and urine N proportion and, on average, led to 0·7-fold lower storage ammonia emission rates on average. Total storage N loss was simultaneously reduced with the extent depending on urine N proportion of the respective manures. A lower dietary protein content furthermore reduced nitrous oxide emission rates in most manure types but increased methane emission from urine-rich slurry; however, global warming potential (based on trace gas output) of all manures was similar with low and high dietary protein content. In deep litter manure, characterized by the highest C:N ratio, emission rates of total N, ammonia and methane were lowest, whereas nitrous oxide values were intermediate. Substantial emission of nitrous oxide occurred with farmyard manure which also had the highest methane values and, consequently, by far the highest global warming potential. C:N ratio of manure was shown to be suitable to predict total N loss from manure during storage in all manure types whereas urine N proportion and manure pH were only of use with liquid manures.