Pharmaceutical reference standards are carefully made samples of a medicine, or of an impurity, whose exact purity is known, against which a laboratory compares its test results. A laboratory cannot measure "how much paracetamol is in this tablet" directly. Its instrument gives a signal, such as the size of a peak on a chart. To turn that signal into an amount, the laboratory measures a sample of known purity in the same way, and compares the two signals. That known sample is the reference standard.
Reference standards are sold by the bodies that publish a pharmacopoeia, the official book of standards for medicines. The Indian Pharmacopoeia Commission sells them for the Indian Pharmacopoeia, and the US Pharmacopeia sells them for its own standards. Each one is made in a large, very pure batch, tested by several laboratories, and sold in small sealed vials with a certificate that states its purity. Each monograph names the reference standard that its tests must use.
Here is how one is used. The laboratory dissolves an exact weight of the reference standard and runs it through a chromatography instrument. Chromatography separates a mixture into its parts, and each part appears as a peak on a chart. The area under the peak rises in proportion to the amount. Then the laboratory runs a solution made from the tablets in exactly the same way. The ratio of the two peak areas gives the amount of medicine in the tablets.
Consider the quality control laboratory at a tablet plant in Sikkim, testing a batch of paracetamol 500 mg tablets. The analyst dissolves 50.0 mg of the paracetamol reference standard, whose certificate states 99.8% purity. She prepares the tablet solution so that it would also hold 50.0 mg of paracetamol if each tablet contained exactly 500 mg. The reference standard gives a peak area of 1,250,000, and the tablet solution gives 1,234,000. The ratio is 0.987. After correcting for the standard's purity of 99.8%, each tablet contains 98.5% of its labelled 500 mg, or about 492.6 mg. The monograph allows 95% to 105%, so the batch passes this test.
Every step must be recorded under GxP rules: which vial of reference standard was used, its certificate, the weights and the instrument's raw data. The instrument and its software must have been proven to work correctly, through computer system validation. If the reference standard is wrong, or past its expiry date, every result calculated from it is wrong too.
How is a reference standard different from a working standard? Reference standards from a pharmacopoeia are expensive and sold in small amounts. So many laboratories make a larger batch of the medicine themselves, test it once against the official reference standard, and use that batch for their daily tests. This in-house batch is called a working standard.