Purwokerto half-fried tempe fritters — thinly-sliced tempe coated in a seasoned flour batter (mostly wheat with some rice flour, generously studded with scallion and faintly colored with a little turmeric) and flash-fried just until the batter is set but still pale and soft (not crispy-golden — that's the wrong stage). Eaten hot with soy-chili dipping sauce. 'Mendo' means 'undercooked' in Banyumas Javanese — the soft, slightly-floppy texture is the defining feature, not a flaw.
Tempe Mendoan is the signature street snack of Purwokerto and the Banyumas region of Central Java — distinguished from regular tempe goreng (crispy fried tempe) by its half-cooked state. Local Banyumas vendors specifically prefer 'tempe Banyumas' which is a wider, thinner-sliced tempe with a slightly looser bean structure ideal for the technique. The batter must be thick enough to coat but loose enough to remain soft after frying. Eaten hot from the wok with soy sauce + sliced bird's eye chili. Modern Indonesian-American restaurants often over-fry into crispy-golden which loses the defining 'mendo' character.
Tempe Mendoan is bizarre to first-timers expecting a crispy fritter — it's pale, soft, slightly floppy, with a barely-set batter that gives way to soft tempe inside. The first bite reveals why it works: the tempe's nutty fermented soybean flavor is intact (not masked by deep-frying), the batter is light and slightly herbal from green onion + turmeric, and the kecap manis dip adds dark sweet-salty complexity. Eat 2-3 with rice and they become breakfast. Eat 5-6 with beer in the afternoon and they become why people move to Banyumas.
The defining technique is undercooking — frying at 160°C (vs typical 175-180°C) for under 2 minutes per side. This sets the batter via starch gelatinization without browning (no Maillard reaction at this lower temperature), keeping the batter pale and soft. The tempe is sliced wide and thin, so it's just heated through. The result is a fritter where you taste tempe + light batter + dip + green onion separately, rather than a uniformly browned fried snack.
Variations
Banyumas canonical (wider tempe slices, looser batter, only 90 sec frying); Yogyakarta variation tends to fry slightly longer (still pale but firmer); modern crispy mendoan (which purists reject — that's just tempe goreng); breakfast-bowl version uses tempe mendoan strips over coconut rice with sambal kecap; some vendors add minced bird's eye chili directly into the batter for spicy mendoan.
On the Palate
Where Tempe Mendoan sits in the Indonesian flavor cloud
Ingredients
Serves 4
How it's made
8 steps · 15 min active + 10 min waiting
- 14 min
Slice 300g tempe Banyumas (or regular tempe) into wide thin slices — about 8cm × 5cm × 5mm thick. You should have ~8-10 slices.
- 23 min
Make batter: in a bowl whisk 100g rice flour + 50g all-purpose flour + 1/2 tsp turmeric powder + 1/2 tsp coriander powder + 1/2 tsp salt + 1/4 tsp white pepper + 2 garlic cloves (grated) + 200ml cold water. The batter should be the consistency of thin pancake batter.
- 31 min
Add 2 tbsp chopped green onion + 2 chopped scallions to the batter. Stir.
- 44 min
Heat 2cm vegetable oil in a wok to 160°C (medium-low — lower than typical deep-frying).
Watch outHold the oil at a gentle 160°C, cooler than normal frying — a lazy shimmer, not a rolling haze; too hot and the batter browns before the tempe warms through.
- 52 min
Dip 2-3 tempe slices in the batter; let excess drip off. Slide gently into the oil.
- 64 min
Fry 90 seconds per side — the batter should set and turn pale yellow but NOT crispy or deep-golden. The tempe inside should be just heated through.
Watch outPull them while the batter is just set and pale yellow, still soft — the whole point is undercooked and floppy, not crisp and golden.
- 71 min
Remove with a slotted spoon to paper towels. Do not stack (will steam and lose the slightly-crisp edges).
- 82 min
Serve immediately with kecap manis + sliced bird's eye chili + sliced shallot in a small bowl as dipping sauce.
Archive tool suggestions — not reviewed
These legacy suggestions have not been evidence-reviewed and are not presented as required tools.
