The neutrino parameters determined from the solar neutrino data and the anti-neutrino parameters determined from KamLAND reactor experiment are in good agreement with each other. However, the best fit points of the two sets differ from each other by about $10^{-5}$ eV$^2$ in mass-square differenc and by about $2^circ$ in the mixing angle. Future solar neutrino and reactor anti-neutrino experiments are likely to reduce the uncertainties in these measurements. This, in turn, can lead to a signal for CPT violation in terms a non-zero difference between neutrino and anti-neutrino parameters. In this paper, we propose a CPT violating mass matrix which can give rise to the above differences in both mass-squared difference and mixing angle and study the constraints imposed by the data on the parameters of the mass matrix.