The Loess Plateau of East Gansu is a typical area where rainfed farming prevails. To understand effects of long-term fertilization on soil nitrogen in soil profile is something very meaningful to farmers in doing fertilization in this area and to researchers in enriching scientific knowledge of soil nitrogen nutrition as well. From wheat fields of a long-term fertilization experiment that started in 1978 and had six treatments, i.e., no fertilizer (CK), nitrogen only (N), nitrogen and phosphorous (NP), nitrogen and phosphorous combined with straw return (SNP), manure only (M), nitrogen and phosphorous combined with manure (MNP), soil samples were collected during the booting and summer fallow stages of wheat in 2014 for analysis of soil total nitrogen (total-N), soil available nitrogen (ava-N), soil mineral nitrogen (min-N) and composition of min-N. Results showed that treatments M and MNP obviously improved soil nitrogen nutrition and were significantly higher than the other treatments in total-N, ava-N, min-N in the 0～40 cm soil layers. No big difference in soil total-N was found between the two treatments. They both increased soil total N by 26.2% on averagely in the 0～20 cm soil layer over the original value in the year of 1978 when the experiment started. Treatment SNP was higher in total-N and ava-N than CK but significantly lower than treatments M and MNP. However, total-N in treatment SNP remained almost the same as or was just slightly higher than the original value. Though treatment NP was quite low in total-N and ava-N, it was still higher than treatments N and CK. Treatments N and NP were 13.1% and 6.4%, respectively, lower than the original value in total-N of the 0~20 cm soil layer. The treatments did not vary much in absolute value of NH₄⁺-N concentration in the soil profiles, but treatment N stood out in NO₃^--N concentration throughout the 0～100 cm soil profile, while the other treatments were higher in the topsoil and sub-topsoil layers than in the 40～100 cm or 60～100 cm soil layer. All the fertilization treatments, particularly treatments M and MNP, lowered NH₄⁺-N/ NO₃^--N ratio in the topsoil layer. Treatment SNP tended to increase the proportion of NH₄⁺-N in soil total N and hence NH₄⁺-N/ NO₃^--N ratio in the middle and lower soil layers. The phenomenon was getting more significant with soil depth in the profile during the summer fallow season. To sum up, under rainfed farming in the Loess Plateau of East Gansu, the practices of applying NP in combination with organic manures are superior to the others in building up soil fertility. The application of N or NP only fails to maintain balance of total-N in the soil. The application of chemical nitrogen alone aggravates downward leaching of NO₃^--N, while the practices of manure, NP and NP plus manure application are proved to be able to reduce NO₃^--N leaching. However, NH₄⁺-N is not much affected by fertilization practices. Fertilization changes composition of min-N, but the effects of manure application and straw incorporation differ quite sharply. It is, therefore, suggested that more efforts should be dedicated to the study on fate and transformation of soil nitrogen under rainfed farming in future.