I also use several passes, but slightly different from what MikeW described in his answer.

• First pass: I reject the photos that are out of focus, have undesired motion blur, wrong exposure, etc., and, at the same time, pick photos which look great. At this state, I don't zoom in to look at details. When shooting in burst, I skip the range of similar photos for the second pass.

• Second pass: I compare similar photos to pick one and reject every other. If I took three photos in burst of a tree in early morning with no tripod, I'll zoom in and choose one which has less hand shake blur compared to other photos of the series.

• Third pass: looping quickly through the unflagged photos, I decide whether the photo has some value or is just a waste of disk space, even if it is done correctly technically speaking. If I like the photo, I pick it. If I don't like it, I reject it. If I'm not sure, I pass to the next photo, but return to this one later.

At this state, I must have every photo flagged pick/reject. Then, I remove the rejected photos and unflag the remaining photos.

• Fourth pass: I loop through the photos, zooming in to see the details. The blurry photos not rejected by mistake during the first pass will be rejected now.

The fourth one must be the last pass, and not be merged with other passes, since zooming in takes much more time. The less photos you zoom in, the fastest will be the process.

On the other hand, you can merge it with Develop workflow (this is what I do most of the times, especially if I'm too excited to adjust the photos I've taken). In this case, either you adjust the photo, or you simply remove it (or reject it, if you're unsure and want to review before removing the rejected photos).