Like a lot of hi-fi gear at the time, these old Marantz receivers were were conservatively rated.
Here's the circuit:
And here's the power supply:
Critical caps: C713, C816 - replace with 50-63 V parts. C403/404 would have to basically go open to affect noise levels severely, but are worth a check.
Zener H810 could use something like 220 µF in parallel, at which point I would also reduce its current to make the filtering more effective by upping R808 to 10-15k (readjust output voltage afterwards).
Critical transistors: The two 2SC458s are arguably the most dodgy candidates, even though 2SC1344 and VD1212 are no unknowns when it comes to intermittency.
Static bursts do not equal permanent hiss though.
If I have my math right, the input transistor is running at a measly 15 µA, possibly less - I've seen <100 µA, even 30 µA, but rarely something quite this low. Surprisingly, this should still yield voltage noise low enough (well below 10 nV/sqrt(Hz)), though it makes input capacitance rather critical (having Vbe of only ~1.7 V does not help) and obviously is rather demanding in terms of transistor leakage. With good parts this circuit should not be unduly noisy at all.
Obviously I would prefer the more common kind of 3-transistor circuit that puts the buffer at the output rather than in the middle (see e.g. Kenwood KA-6004). That said, this circuit here seems to have been used in quite a range of vintage Marantz receivers for several years. Can't have been
that bad.