Hi, I attach our old publication and Guidance of the Eurachem 2014. Is your compound stable in the selected solvent? Is your target peak pure (not contaminated)? If your target peak is not stable and not pure, it is possible the peaks did not yield height or area as expected. We usually used peak area instead of peak height.  I think it is not so nice to omit one/some points in the calibration standards.  Let me know if you have some more questions.

I am sorry to hear you are having trouble with your method. It can be frustrating when learning a new technique, but the solutions can be found if you start with the fundamentals, make sure everything is working properly and most of all, that things make scientific sense.

Please do not omit any peaks from your calibration table. Sounds very unscientific indeed. Instead, determine why your calibration data is unreliable and not reproducible. What is causing it? It is trying to tell you something. Take the hint and listen to the clues provided. Any number of method development flaws, problems or poor technique could cause what you observe. Additionally, there is a long list of possible instrument (hardware) issues that could also result in poor reproducibility too. 

- So, where to start troubleshooting? Here are a few ideas:

• First, clearly define the problem and its frequency. Is it the same std each time? What are the conditions when it happens? Compare runs.
• Sample Prep: You mentioned that you are performing serial dilutions. These are often done with high error. Can you change this method or use an automated injector to create the different dilutions?
• Column Temperature must be stable throughout the day (+/- 1C) or retention times may drift and/or RSD may be high.
• If possible,  use peak areas, not heights in HPLC. They tend to be more reliable and are the accepted std in HPLC.
• Make sure your baseline is as flat and stable as possible during the analysis.
• Use consistent integration parameters and check the system to make sure it is making logical choices where peaks start and stop to calculate areas.
• Check your injector's rotor seal and stator for contamination or wear (these require regular maintenance. A worn or contaminated injector will result in poor RSD).
• Make sure your analysis method is followed by a proper wash method to insure that all material is washed off the column. It should incorporate a mobile phase mix which is stronger than the one in your analysis method. A method which does retain, but not elute ALL material off the column in one run will store some of the samples for the next run, leading to larger peaks and new peaks.
• Use enough cal point stds (min 5) to cover the entire range of possible samples. We do not extrapolate, only interpolate. Select a curve fit which closely approximates the actual data. Run lots of replicates to test.

Lastly, here is a link to an article which discusses common reasons for poor RSD in methods. Some of the items mentioned may also be responsible for your issues too. 

http://hplctips.blogspot.com/2015/11/hplc-retention-time-drift-change.html

Hello Benjamen, Calibration curve is a great way to detect problems in the standards. Since you are getting odd results, it means there is something wrong either in your standards (very likely) or the instrument (last one to be blamed). The best way to solve problems is to take baby steps in cross-checking the whole process right from the beginning. If you could show the chromatograms, that would be great as well. The most common mistake, among new users, is usually in pipetting and hence in preparing the standards. Are you using air displacement pipettes? You can easily check your /lab mate pipetting technique by pipetting a known volume of water, and weighing the volume delivered on a balance e.g. 500 uL H2O should weigh something close to 0.49 gram, if everything is alright. If this value is significantly off, the pipette must be checked by the instructor. I have seen large errors in this technique. Many students ignore this and thus get incorrect results from pipetting techniques. Practice drawing volumes with couple of other settings. Also recheck the volumes you used in serial dilution (simple cross check again). As others have said, check if the sample is stable in solution. For example, when I did HPLC of acetylsalicylic acid, it would break down into salicylic acid within 24 hours- the result was two peaks instead of one.

HPLCs are pretty robust. I believe you equilibrate the column for some time. Ensure that each time the injected volume is the same. One thing which you/instructor should check, is something called as the sampling frequency (it is present in the detector settings). If it is two low, you can exactly see the same problem in poor reproducibility of peak height. What time does you peak elute? Try 20 Hz setting and see if it helps.