Ask how many people joined the cast in 1985 and the honest answer is a question back: joined, or arrived?
Eleven people, and also twelve
The roster holds 174 records. Eleven of them give 1985 as the first year the person was ever credited on the show. Twelve of them contain a run that begins in 1985. Both counts are correct. Nothing is missing and nothing has been counted twice; the two figures answer different questions that sound identical in English.
The twelfth is Don Novello. His record carries two separate runs, 1979–1980 and 1985–1986. If you are counting the people for whom 1985 began a stretch at Saturday Night Live, he is not one of them: he had already been and gone. If you are counting the runs that started in 1985, his second one did. The twelve are Joan Cusack, Robert Downey Jr., Nora Dunn, Anthony Michael Hall, Jon Lovitz, Dennis Miller, Don Novello, Randy Quaid, Terry Sweeney, Danitra Vance, Dan Vitale and Damon Wayans; the eleven are that list with Novello taken out of it. Neither sentence is a rounding of the other.
A third figure exists for that year, and it is the one most readers have in mind. Twenty-two people were credited at some point during 1985, because ten had arrived earlier and were still there. Harry Shearer is one of the ten: his second run reads 1984–1985, so he was in the cast during 1985 without starting anything in it. Eleven, twelve, twenty-two. One year, no errors. The difference is which question you asked.
What makes this a trap rather than trivia is how rare the trigger is. Only four of the 174 people have more than one run: Brian Doyle-Murray, Al Franken, Don Novello and Harry Shearer. Franken has three, the others two, and that is the whole list. In total the roster holds 179 runs for 174 people, a gap of five. So the two counts agree for most years and then diverge for the handful that matter, and nothing warns you when they do.
A second seam runs alongside the first. Thirteen records contain a run that starts and ends in the same year: George Coe and Michael O'Donoghue in 1975, four more in 1980, two in 1981, and Ben Stiller, Morwenna Banks and Fred Wolf in 1989, 1995 and 1996. Eleven of the thirteen are people whose entire time on the show was that one run; the other two, Doyle-Murray and Franken, have a same-year run sitting inside a longer record. Any arithmetic that subtracts a start year from an end year reads all thirteen as zero, so every rule needs a special case for them.
The tempting response is to pick one definition, call the others sloppy, and move on. That is the wrong instinct. First arrivals and run starts are not competing attempts at one measurement; they measure different things, careers begun and shifts begun. A piece about who the class of 1985 was wants the eleven. A piece about how much new blood walked through the door that autumn wants the twelve, because Novello walking back in was as much a hire as anyone else's first day. The mistake is never picking the wrong one. It is picking one silently and letting the reader assume you picked the other.
The harder case is seasons
Years are the easy half, because a year is at least a fixed thing. Seasons are the hard half, because a season of this show is not a year: it opens in the autumn of one calendar year and closes in the spring of the next, and people are hired and dropped in the middle of it. Every record carries a season count that was sourced rather than calculated, and the point of that number is that it does not have to agree with arithmetic on the year pair.
Test the obvious rule and watch it fail. Take the inclusive span of each run, end year minus start year plus one, add them together, and treat anyone still on the show as running to 2027, the year season 52 ends. That rule disagrees with the sourced season count for 142 of the 174 records and agrees for 32. Something that gets more than four records in five wrong is not a rule with exceptions. It has misunderstood what it is counting.
What it has misunderstood is the autumn boundary. A record reading 1980 to 1984 does not describe five seasons. It describes someone who joined during the 1980–81 season and left after the 1983–84 one, which is four. Eddie Murphy is exactly that record: his run reads 1980–1984 and his season count is 4. Subtracting the years gives 4 and matches; adding one gives 5 and does not. So stop adding the one, and keep a floor of one season so the thirteen same-year runs do not collapse to zero. Under that corrected rule the disagreement falls from 142 records to 25, and 149 of the 174 come out right.
Twenty-five beats 142 but is not zero, and the residue is the part worth reading. In 24 of the 25 the sourced count is higher than the rule, which is what mid-season hiring does to a record. Chevy Chase is the plainest case: his run reads 1975–1976 and his season count is 2, where the rule says 1. He was in the first season and part of the second, which the arithmetic cannot see. Maya Rudolph is the widest gap of this kind on the roster, and the only one worth two whole seasons: a run of 2000–2007 against a season count of 9 where the rule says 7. Nobody else beats the rule by more than one. Fred Wolf makes the same point in miniature, a run beginning and ending inside 1996 and a season count of 2, because one calendar year can hold the back half of one season and the front half of the next.
Exactly one record runs the other way. Al Franken's three runs total 9 under the corrected rule against a sourced count of 8, because the rule charges him a full season for a run recorded as 1986 to 1986. One record in 174 pointing the other way matters: the mismatches are not a systematic offset you could subtract away. Current cast members add their own wrinkle, since their run has no end year and the arithmetic must substitute one. Colin Jost's run reads 2014 to the present, the rule reads 13 seasons, and his sourced count is 14. Which is right depends on where in a season he started and whether a season in progress counts, and the record does not say.
The defensible position, and one a reader may argue with, is that the sourced count should never be overwritten to match the arithmetic. The counter-argument is real: a dataset in which 142 records fail your own formula looks broken, and a site could buy perfect consistency in an afternoon by recomputing the column. It would be the wrong purchase. The formula guesses where season boundaries fell; the sourced count records seasons somebody appeared in. Replacing an observation with an estimate is not accuracy, it is tidiness, and it erases precisely the 25 records that had something to teach you.
The same shape of problem sits in the role flags. Each record carries seven, from repertory and featured through to Weekend Update, hosted and writer, and every one is a plain true or false with no date near it. A writer flag set to true means the person was credited as a writer at some point in their association with the show. It does not say during which run, for how many seasons, or alongside what. The hosting and Update flags behave identically. They are associations without dates, which is the same failure as eleven against twelve: the data answers a narrower question than the one a reader assumes.
None of this argues for giving up on counting. It argues for the count and the rule travelling together. "Twelve people joined in 1985" is worth little alone and worth a lot as "twelve runs began in 1985, of which eleven were first arrivals". The rules this site uses, and the reasons for each, are set out in a companion piece on the methodology behind these counts. The raw runs are on the roster and in the individual biographies if you want to check the working, and more counting of this kind is on the articles page.
Cast dates and tenures on this site come from Wikipedia; the counting and the reading of it above are our own. More of these are on the articles page, and corrections are welcome at info@snlroster.com.