While attempts to promote acceptance of well-evidenced science have historically focused on increasing scientific knowledge, it is now thought that for acceptance of science, trust in, rather than simply knowledge of, science is foundational. Here we employ the COVID-19 pandemic as a natural experiment on trust modulation as it has enabled unprecedented exposure of science. We ask whether trust in science has on the average altered, whether trust has changed the same way for all and, if people have responded differently, what predicts these differences? We 1) categorize the nature of self-reported change in trust in “scientists” in a random sample of over 2000 UK adults after the introduction of the first COVID vaccines, 2) ask whether any reported change is likely to be real through consideration of both a negative control and through experiment, and 3) address what predicts change in trust considering sex, educational attainment, religiosity, political attitude, age and pre-pandemic reported trust. We find that many more (33%) report increased trust towards “scientists” than report decreased trust (7%), effects of this magnitude not being seen in negative controls. Only age and prior degree of trust predict change in trust, the older population increasing trust more. The prior degree of trust effect is such that those who say they did not trust science prior to the pandemic are more likely to report becoming less trusting, indicative of both trust polarization and a backfire effect. Since change in trust is predictive of willingness to have a COVID-19 vaccine, it is likely that these changes have public health consequences.