1: \begin{abstract} 

2: We revisit the method of small subgraph conditioning, used to

3: establish that random regular graphs are Hamiltonian a.a.s. We refine

4: this method using new technical machinery 

5: for random

6: $d$-regular graphs on $n$ vertices that hold not just asymptotically,

7: but for any values of $d$ and $n$. This lets us estimate how quickly

8: the probability of containing a Hamiltonian cycle converges to $1$,

9: and it produces quantitative contiguity results between different

10: models of random regular graphs. These results hold with $d$ held

11: fixed or growing to infinity with $n$. As additional applications, we establish

12: the distributional convergence of the number of Hamiltonian cycles

13: when $d$ grows slowly to infinity, and we prove that the number of

14: Hamiltonian cycles can be approximately computed from the graph's

15: eigenvalues for almost all regular graphs.\end{abstract}

16: