The most common cause of peak drift that I've seen is due to pH change of your solution. If the pH is close to the pKa of your analyte only slight changes in pH are needed to shift the peak. To avoid this make sure you use a buffer solution that keeps the mobile phase at least 2 pH units away from the pKa of your analyte.
Other possible reasons can be temperature - is your column thermostatically controlled? Column blockage - make sure you flush the column thoroughly after the run and, eliminate the obvious, has anyone else been using the machine that may have altered the method without you knowing?
It seems that you have a problem with the first column, second transient or unstable pH. The third reason may be the same way injection. If injection is made automatically by the autosampler, it remains to establish a good pH after rinsing the column. Note the temperature!