We use a single quad MS detector coupled to a TSP P4000 HPLC pump. Using adequate mobile phases in gradient elution (water + TFA, ACN), we got very deconcerting results on our system. The method was first developed and optimized on another LC-UV system, that yielded us beautiful chromatograms and resolution. We developed the method with both a 4.6 mm id column and a 2 mm id column : the results were systematically as expected on the LC-UV system. Then we transferred the column (2 mm, flow 0.5 ml/min) on the LC-MS : the first injection yielded good results, but the second one and all the following gave us reproducible bad results : peaks coming too quickly (exactly 2 times quicker), no more resolution. We tried the 4.6 mm column (flow 1 ml/min) and got expected, reproducible good results. We did exactly the same with another molecule and another gradient : irreproducible at lower flows (second injection coming two times quicker, no resolution), good and reproducible at higher flows. We suspected a problem with the pump ; we added an UV detector and set-up a 0.1% aceton gradient according to our SOPs. The pump yielded excellent results at both 1 ml/min and 0.3 ml/min. What could be the cause of these deconcerting results ?