Is it possible that carryover is causing your peak amounts to increase? Do you notice any residual sample if you run a blank? You can try weighing your vial after each injection to ensure that the sampler is drawing the same amount of liquid each time.
If you use a sample which have a high viscosity, then it helps to increase the injection time.
You can also try to inject 5 Blank runs to see if you see a change in the baseline when u overlay the signals.
You can also try this step with your samples.
You also might control your pressure / flow.
When you say that there are no signs of a leak, have you run a pump leak rate test using Agilent Lab Advisor? If not, I would suggest testing this.
If you're unlucky enough for your peak to elute just at the point when the outlet ball valve fails to close properly, the flow will reduce momentarily be reduced, your peak will stay in the flow cell for longer than it would at full flow and the peak will have a larger area than expected!
Does that make sense?