The 7 Sedans That Hold Their Value Best in Malaysia

01The seven sedans we can stand behind

The sedan has quietly stopped being Malaysia's default car and become its most misunderstood one. It is also the segment where our two independent methods argue with each other most often — so this table is shorter than the others in this series, and the reasons for that are worth more than the ranking itself.

02What each row costs to run

Sedans are assessed on Malaysia's cheaper road-tax scale, and in the volume part of this table the annual bill is close to trivial. That is a genuine structural advantage over the equivalent SUV.

Malaysia runs two road-tax schedules: one for saloons and one for SUVs, MPVs, vans and pickups. Below roughly 2,300cc the saloon scale is the cheaper of the two, and the sedans here sit comfortably under that line. A 1,500cc sedan pays RM90 a year where a 1,500cc SUV pays RM120, and a 1,332cc Proton Saga pays RM70. The rates come from JPJ's licence-rate calculation guideline; the SUV side of the comparison is in our SUV retention table.

03Malaysia's best-selling car is a sedan we refuse to rank

The Perodua Bezza was the best-selling vehicle in the country in 2025 with 100,488 units registered, its third consecutive year at the top. The Proton Saga took fourth with 70,903, the Toyota Vios seventh with 31,776 and the Honda City ninth with 27,107. Three of those four are in the table above. The biggest one is not.

The Bezza produced a curve. It has 3,156 listings across 97 model-year cohorts and a ten-year span — more depth than most rows we did publish. It failed on quality, in two separate ways at once.

04Why the sedan is the hardest segment to measure

Every article in this series ends up shorter than the data first suggests. The sedan article is the shortest, and the pattern behind that is consistent enough to publish.

Sedans clear the depth and fit-quality gates as readily as anything else — 15 of them did, the same number as SUVs. They then fail the corroboration step at roughly twice the rate. Six sedans were dropped as contested, and those six are among the best-known nameplates in the country.

The disagreement runs one way, and we cannot yet explain it

Across the whole market, nine models were excluded as contested. In every one of the nine, the fitted curve reads lower than the observed price ladder — not one runs the other way. Among the thirty models where the two methods do agree, the direction is mixed: twenty-one lean one way, nine the other.

We tested the obvious explanation — that the contested models carry longer histories, so a single curve is being stretched across more model generations — and the data does not support it. The contested sedans have a median span of 20.5 years against 23.0 for the corroborated ones, which is the wrong way round. The Honda Civic and the Honda City have almost identical sample depth and span, and land on opposite sides of the line.

So we are reporting a property of our own estimator that we have not yet accounted for, rather than a finding about sedans. It is the reason the corroboration step exists. When the two methods part company we publish neither number, and we say which models that happened to.

One more measurement note

The Mazda 3 does not appear here at all, and not because it failed a retention gate — it is corroborated in our market-wide table at 67.1%. It fails a different test: the nameplate is sold in Malaysia as both a sedan and a hatchback, and the body type recorded against its listings splits too evenly for us to call the cohort one or the other. Where a nameplate spans two body styles, we leave it out of the body-style tables rather than assign it arbitrarily.

05What to actually do with this

Seven rows is a small table, but the decisions it supports are the common ones.