Since 1972, the General Social Survey has asked "Generally speaking, would you say that most people can be trusted or that you can't be too careful in dealing with people?" The question has been widely used in research, especially after the publication Robert Putnam's Bowling Alone, and it's been picked up by other survey organizations. Since 2010, it's been asked in 27 different national surveys.
There's another question on trust that's not nearly as well known. Until recently, I either didn't know about it or had forgotten about it. It is "Do you think most people can be trusted?" It was first asked in 1942, and then asked off and on until 1964. After that it has been asked only once, when the GSS included it as an alternate form of its usual trust question. The percentages answering "yes":
The lowest level of agreement was in 1983 (56%), but it was almost as low in November 1953 (57%). The samples were not all that large, so you could probably make a case that the apparent differences between surveys in the early 1950s were mostly (or maybe all) sampling error. For the same reason, the series doesn't tell us much about long-term change, since there's only one post-1960s observation (I'm not certain of sample size, but I think about 800).
There's one thing that would let us say more about long-term change, which would be if someone would ask the question today. I'm surprised that hasn't been done.
[Data from the Roper Center for Public Opinion Research]