Daily drill cards in med school, written as a clock, not a feature tour.

The pattern that holds up across an MS1 block looks like this: convert the day's lecture deck to questions the same evening it was given (about 60 seconds of setup), drill 30 to 60 questions from that single deck for roughly 5 minutes that night, then let a spaced-repetition scheduler interleave the older decks across the rest of the week. The deliberate part is that each session is anchored to one specific deck, not a global queue. A global queue of 'all cards due today' grows scary fast in med school because new cards keep arriving on their own schedule on top of the back-half of last week's lectures. A per-deck nightly anchor keeps the session short and keeps the failure mode local: skip one night and one deck's review slips, not all of them.

Both, from the same source, drilled in one mixed pass. A named clinical sign is at least four facets (the maneuver, the condition, the mechanism, the look-alike sibling sign); a flat term-to-definition flashcard tests one of them. A multiple-choice stem tests recognition under distractor pressure, which is closer to what the exam grades. The honest answer is to upload one deck and ask the tool to produce both formats, then alternate them inside a single 5-minute pass: a few MCQs to force recognition under a wrong-answer trap, a few free-response or image-occlusion cards to force unprompted recall. The point is not picking a format religiously; it is making sure each fact gets retrieved more than one way before exam week.

If you convert each lecture deck the day it is given, a typical 90-slide block lecture yields 60 to 80 generated multiple-choice questions plus 20 to 30 image-occlusion or free-response cards. You will not drill all of them on day one. The viable pace is 30 to 60 retrieval attempts per nightly session, mixing today's new deck with whatever cards the scheduler resurfaces from the last 5 to 10 days. Pushing past 60 attempts per session is where people start consolidating sessions into 'I will do a long block on Sunday', which is the exact pattern that collapses by week 3.

Two failure modes, only one of which is real. The fake failure mode is a broken streak: a global counter resets to zero, you feel demoralised, you skip again. Anki's maintainer resists shipping a native streak for exactly this reason, recorded in GitHub issue #4085: an all-or-nothing chain pushes people into clock-backdating and one-card sessions just to keep the number alive. The real failure mode is a backlog: the questions you owed yesterday stack on top of today, plus the new deck from today's lecture. The fix that works is per-deck progress (each deck has its own tree, a missed day pauses one tree at its current height, the rest of your decks keep their progress) and lowering new-card intake for a day or two until the queue is back inside your nightly time budget. Catastrophising a single skipped day is a worse strategy than absorbing it.

Anchor it to something you already do every day without deciding. For most med students that is one of three slots: between the last lecture and dinner (works only if your block has clean afternoon endings), the 10 minutes after you close the laptop on the day's note-taking, or the last 5 minutes before bed. The specific slot matters less than the anchor: a session decided fresh every day stops happening by week two. The other rule is to do the session BEFORE switching to a longer 'I should also do qbank questions' block. The daily drill is the load-bearing minimum; the longer sessions are optional on top.

Daily MS1 drilling is the foundation for surviving MS2 and dedicated, not a substitute for them. The catch is what you drill on. Most MS1 block exams are professor-written from specific slide decks, not Step 1 blueprinted, so the highest-yield daily card during MS1 is one generated from your professor's lecture, not from a board question bank. Save UWorld for dedicated; spend MS1 nightly time on the deck your block exam will actually pull from. The transfer to Step 1 happens because the underlying physiology, pharm, and path do not change between a slide-fact question and a clinical vignette: a retrieval rep on potassium handling is a retrieval rep either way.

Anki the scheduler is fine and you can run this protocol entirely inside Anki if you want. The catch is the authoring step. Hand-making cards from a 90-slide deck runs 60 to 120 minutes for most people; do that for 4 decks in a day and the authoring tax alone is longer than lecture itself. Most people who 'fall off Anki' did not fall off spaced repetition, they fell off the cost of feeding it. The other gap is on the reward side: Anki shows one number (cards due) that resets to zero each session, so showing up leaves no visible mark and skipping doubles tomorrow's queue. A per-deck visual that accumulates with each session, plus auto-rephrasing on revisit so a second pass through the deck is a real retrieval and not a sentence-shape match, are the two pieces missing from a default Anki setup.

Five minutes a night per deck, drilling cards generated against that specific deck, is enough to keep the retrieval curve flat between the day the lecture was given and the day of the block exam. It is NOT enough as the only thing you do; lecture, notes, and any case-based work still sit on top of it. The reason 5 minutes works is density. Drill cards generated from your own lecture have a much higher exam-relevance hit rate than the same time spent on a generic web bank, because the question source matches the test source. In weeks where the block lands 4 decks on the same day (a cardiology week, for example), 5 minutes becomes 20 minutes split across the 4 trees, not a 4x longer single session.