P-value is the calculated probability of finding the observed variation when the null hypothesis (Ho) is true. It can also be understood as the calculated probability of committing a Type I error (i.e. accepting alternative hypothesis when actually null hypothesis is true). The outcome of the statistical test (to accept null or alternative hypothesis) is determined basis the comparison of p-value with significance level.
An application oriented question on the topic along with responses can be seen below. The best answer was provided by R Rajesh on 24th September 2018.