Field experiments with fixed crop rotation of finger millet-maize-cowpea sequence were conducted in the on-going All India Co-ordinated Research Project on Long Term Fertilizer Experiment at Tamil Nadu Agricultural University farm, Coimbatore since 1972. The treatments are: T1:50% NPK; T2:100% NPK (optimal); T3:150% NPK; T4:100% NPK with hand weeding, T5:100% NPK + ZnSO4 (maize); T6:100% NP; T7:100% N alone; T8:100% NPK+FYM (finger millet); T9:100% NPK (S free) and T10:Control. In the 100 per cent NPK treatments, a total of 6250 kg N ha-1 has been added through chemical fertilizer viz., urea as 100 per cent N to finger millet, maize and grain cowpea (both grain and fodder) crops during the years from 1972-73 to 2002 - 2003. Total removal of N was maximum 7662 kg N ha-1 with the application of 100 per cent NPK + FYM where the per cent recovery was 93. However, the maximum N recovery was obtained with 50% NPK. With the application of 150 % NPK, the overall recovery was 74. The lowest N recovery was observed with 100% N alone treatment. In the 100 per cent NPK treatments, a total of 6250 kg N ha-1 has been added through chemical fertilizer viz., urea as 100 per cent N to finger millet, maize and cowpea (both grain and fodder) crops during the years 1972-73 to 2002-2003. Total removal of N was maximum 7662 kg N ha-1 with the application of 100 per cent NPK + FYM where the per cent recovery was 93. However, the maximum N recovery was obtained with 50% NPK. With the application of 150 % NPK, the overall recovery was 74. The lowest N recovery was observed with 100% N alone treatment. The highest positive balance of 3449 kg N ha-1 was evident in 100 per cent N alone treatment, which might be due to poor dry matter yield and uptake by crops. Compared to the initial status of 178 kg ha-1, the available N status was not drastically reduced in any of the treatments including the control plot. This indicates that the soil had the capacity to replenish the available N status by recycling the crop residues. The available N status was higher in 100% NPK+FYM and in 150% NPK plots and it was lower in the control. This indicates that optimal or over and above the optimal doses of NPK or application with organic manures may have to be resorted to for maintaining the soil available N status over the years. Continuous exclusion of fertilizers or disproportionate application of NPK such as N or NP alone or below optimal doses such as 50% NPK may not enrich soil N due to reduced crop yields and hence, low return of crop residues for recycling of organic matter in the soil. The total P ha-1 added under 100 per cent NPK application to all the crops during the years 1972 - 73 to 2002 - 03 was 1773 kg. Its removal ranged from 225 to 1082 kg P ha-1. The available P status has increased to about 15.1 to 17.7 18.0 to 18.7 kg ha-1 in 150% NPK and 100 per cent NPK + FYM plots respectively from the initial status of 11.0 kg ha-1. The results clearly showed that the Inceptisol get depleted to the extent of 288 kg P ha-1 during the past 32 years when N alone is applied over a long run. Similarly in control plot, the depletion was found to be 225 kg P ha-1. Wherever P is applied, the balance is positive ranging from 209 to 1713 kg P ha-1. The highest positive balance of about 1713 kg P ha-1 was observed under 150 per cent NPK treatment under intensive cropping. There was a reduction in P availability in all the treatments except 150% NPK and 100% NPK+FYM plots during the last thirty two years. As the applied P was continuously utilized by the growing crops and the soils being only medium in available P, there was a reduction in P availability in most of the treatments. The magnitude of reduction in P availability was more in 50% NPK, 100% N and control plots. The magnitude of P depletion was more wherever it was not applied or applied below optimal amounts. The available P status was maintained on par with or slightly above the initial levels at the start of the experiment in 1972 wherever it was applied above optimal levels (150% NPK) or in combination with organic manures (100% NPK+FYM). The total K balance in all the input systems is negative unlike N and P. Also, the total nutrient removal by the crops in the sequence was much higher including the control than the added dose of K fertilizers. The maximum K removal of 8731 kg K ha-1 was by 100% NPK + FYM treatment as against the addition of 2750 kg ha-1. The 150 per cent NPK treatment removed about 7976 kg K ha-1 over 32 years as against addition of K of 1634 kg K ha-1. The highest negative balance of 6411 kg K ha-1 was observed with 100 per cent NP treatment, which might be due to imbalanced fertilizer application. The absence of K fertilization lead to about 2216 to 6411 kg K ha-1 mining of K from the native reserve of soil. The available K was significantly lower in control and 100% N treatments while 100% NPK + FYM and 150% NPK treatment recorded higher available K over the years. Graded levels of NPK from 50 to 100% did not raise the availability of K in the soil to any significant effect but further increase in the levels of NPK (150% NPK) had a positive effect on the K balance in the soil. The continuous depletion of K irrespective of its application or not, indicates that K is continuously being mined from the soil. Compared to the initial level of 810 kg/ha, there was a depletion of available K in all the treatment irrespective of whether K was applied or not. Key words: Long term fertilization, nutrient balance and availability.