Well, let's see.
You've got a series resonant tank circuit. C = 1u, L = 8.2m, R = 1. V(0) = 350, I(0) = 0.
The parameters are:
Resonant impedance Zo = sqrt(L/C) = sqrt(8200), about 90 ohms.
Resonant frequency, well I don't really care because it's on a scaled horizontal axis anyway. But that's Fo = 1 / (2*pi*sqrt(L*C)) if you like. (The exact frequency of zero crossings is the pseudofrequency: it's not a periodic system, because the amplitude is decaying between zeroes. Usually, this has a damping term that decreases the frequency slightly.)
Q factor Zo / R = 90, which means it will decay by e in about 90 cycles. Which... uh... hmm.
You didn't change the defaults, did you?
Yeah, LTSpice does that...
In fact, it's decayed that far in about 2.5 cycles, so the ESR + DCR total is around 36 ohms. (Seems awfully high for defaults, but there may be a parallel equivalent instead.)
Anyway, since the resonant impedance is 90 ohms, we know the first quarter wave has to be a current maxima, where Ipk <= Vpk / Zo, or 3.89A.
It's actually more like 3.5A on the first peak, another indicator of circuit losses.
So the damping is wrong because of parasitics (and possibly numerical stability; try GEAR solver and RELTOL = 0.0001), but the peak values, early on, are consistent.
These calculators may be of relevance, too:
http://seventransistorlabs.com/Calc/RLC.htmlTim