Add a fixed amount of weight every session, on every lift, for as long as that keeps working. 3 sessions a week, two alternating workouts.
Just starting out? Read this page top to bottom before your first session — it is short and it is the whole program. Learn the four lifts first
Already stalling on this program? Check the stall and reset rules below, then see if intermediate is the right move. Stalls and resets · Intermediate phase
Two workouts, alternated, 3 times a week — never the same workout twice in a row, and never two sessions on consecutive days. A typical week is A, B, A; the next week starts on B.
| Day | Lifts | Sets × reps |
|---|---|---|
| Day A | Squat, bench press, deadlift | 3 × 5, 3 × 5, 1 × 5 |
| Day B | Squat, overhead press, deadlift | 3 × 5, 3 × 5, 1 × 5 |
Squat and deadlift train every single session, on both days — a deliberate choice for frequency over the "deadlift once a week" pattern some programming favors, because practicing the pattern matters more than the extra recovery time would this early on. Bench press and overhead press alternate, one per day, so each still trains roughly every other session across a normal week.
Start with the empty bar on every lift and let the fixed per-session increments carry you up from there — this is the default the app itself uses at enrollment if you don't supply a number. If you already know a comfortable working weight for a lift, you can enter that instead and start from there. Either way, the first few sessions are about grooving the pattern, not testing a limit — an empty-bar or very light week is a completely normal, correct way to begin. See the finding your starting numbers page for why starting too heavy backfires specifically in this program, and a real method for picking a number without a maxing-out session.
A session "succeeds" for a lift when you complete every prescribed rep on every prescribed set. When it does, that lift's working weight goes up by a small, fixed amount next time you train it — no judgment call, no percentage math, the same number every time, in whichever unit your own plates come in.
| Lift | Sets × reps | Add per successful session, from your first miss on |
|---|---|---|
| Squat | 3 × 5 | +5 lb (2.5 kg) |
| Bench press | 3 × 5 | +5 lb (2.5 kg) |
| Overhead press | 3 × 5 | +5 lb (2.5 kg) |
| Deadlift | 1 × 5 | +10 lb (5 kg) |
Deadlift takes the biggest jump because its hip-hinge pattern involves the most total muscle mass; bench and press take the smallest, matching how much more slowly upper-body pressing strength typically builds for a new lifter. These are the real Starting Strength figures — this program follows them directly rather than inventing its own.
Your very first sessions on each lift add weight faster than the table above: double the steady rate — +10 lb (5 kg) on squat, bench, and press, +20 lb (10 kg) on deadlift. That faster jump is only safe to trust while your true working weight is still unknown and the bar is trivially light for you — the instant you miss a single rep on a lift, even the first time it ever happens, that lift permanently drops to its steady rate from the table above. It never speeds back up after that.
There's a real theoretical reason a novice can add weight this often when later phases can't — see why the clock changes for the actual adaptation-timeline research behind it, not just this program's own say-so.
Adding weight almost every session, for months, is a real recovery demand — see nutrition basics for this program for why that changes what you actually need from food compared to general fitness advice, and how to get real numbers instead of guessing, sleep and recovery basics for real sleep-duration guidance and what the rest days below are actually for, accessory work for a novice for why that same recovery demand is also the real reason to add little or nothing beyond these four lifts for now, and conditioning and cardio alongside the program for whether running or biking on the side will get in the way.
A lift "stalls" when you fail to complete every prescribed rep, at the same weight, for 2 sessions in a row — not the first missed rep, but the second consecutive full session of missing. When that happens, the lift resets: its working weight drops to 90% of the last weight you actually completed in full, and you build back up from there with the same fixed increment.
Two more rules can reduce the increment itself, and both can apply to the same reset:
| Rule | What it does | Applies to |
|---|---|---|
| Press step-down | Overhead press's own increment drops from +5 lb (2.5 kg) to +2.5 lb (1.25 kg) the moment it resets for the first time | Overhead press only |
| Second-reset halving | Any lift's current increment is cut in half if it resets a second time within a rolling 28-day window | Every lift |
The press steps down first because it's one of the two lifts Starting Strength's own coaching material names as usually the first to stall in a novice's progression — its article "Don't Jump Ship: Earning the Transition from Novice to Intermediate Training" explains why: the press and bench press "are often the first lifts to stall, because even though they aren't as systemically stressful as squats and deadlifts, they use much less muscle mass and therefore have less contractile mass contributing to the lifts." This program's own step-down rule targets the press specifically, not bench, since the press runs meaningfully lighter and weaker than bench for most novices — see why overhead press runs lighter than bench for the real data behind that gap. The second-reset rule above is general, not press-specific, because repeated resets on any lift in a short window are the actual signal that the daily-progression model itself is running out, not a fluke week.
Many women hit that stall-and-reset point, and need a smaller increment, earlier in the progression than many men do — a well-known pattern in novice lifters, not a sign of doing the program wrong. The app doesn't run a separate rule for this: the same reset and halving logic above responds to real stalls whenever they actually happen, so a lifter who needs to downshift sooner gets a smaller jump automatically, without anyone having to ask for it. It's also why fractional plates matter especially for women, and usually sooner — every novice lifter benefits from them eventually, men included, just typically later. See the equipment page for the actual plates, and older, younger, and women lifters for real training differences for women beyond this one rule.
Two mistakes with this rule aren't about the numbers at all: rounding a grindy, form-breaking rep up to a success so a stall never gets counted, and refusing an earned reset by hand-entering a heavier number anyway. Both are covered on the common early mistakes page.
None of this assumes a stall is always about fatigue. If a lift keeps missing because a specific movement genuinely hurts rather than just feels heavy, that's a different problem with its own real tools — see training around minor aches and injuries for the decision tree.
You don't have to plan warm-up sets by hand. The app computes a ramp from the empty bar up to each session's working weight, using the actual plates in your rack — the same plate-by-plate ramp the rest of the app is built around. On paper, a reasonable approach is a small number of ascending warm-up sets with shrinking rep counts as the weight climbs, finishing your last warm-up meaningfully below your working weight for the day. See the warm-up and plate math page for why this program ramps up at all, the exact ramp the app runs for squat/bench/press versus deadlift, and how to load the bar by hand if you're running this on paper.
Moving off novice is always your own call, never automatic. The app flags a lift as a graduation candidate once it has reset 2 times — a signal to consider moving that lift to intermediate-style handling, not a trigger that moves anything without asking. It's normal for different lifts to reach this point at different times — the app tracks each lift's own reset count independently, and it's common to move a stalling lift onto intermediate-style handling while others are still climbing on the novice model. See the intermediate phase page for what changes.