ValueError: could not convert string to float: '12,5'

ValueError: could not convert string to float: '12,5' means float() received text that is not a valid number. The text after the colon is exactly what Python failed to read, so look at it first: a decimal comma, a thousands separator, a currency sign, a unit, or nothing at all. Clean the text, then convert.

Reproduce it

python3 - <<'PY'
for text in ['12.5', '12,5', '']:
    try:
        print(repr(text), '->', float(text))
    except ValueError as e:
        print(repr(text), '->', type(e).__name__ + ': ' + str(e))
raise SystemExit(1)
PY

'12.5' -> 12.5
'12,5' -> ValueError: could not convert string to float: '12,5'
'' -> ValueError: could not convert string to float: ''

The first value converts. The other two show the typical failures. An empty string gives could not convert string to float: '', which is easy to misread because the quotes look empty.

What float() accepts

Fixes

Decimal comma and thousands separators

Decide which character is the decimal separator for your source, then normalise:

python3 - <<'PY'
print(float('12,5'.replace(',', '.')))
print(float('1,000.5'.replace(',', '')))
print(float('$12.50'.lstrip('$')))
PY

12.5
1000.5
12.5

For European text such as 1.234,5 remove the dot first and then replace the comma: s.replace('.', '').replace(',', '.'). Do not apply that rule to English-format numbers.

Empty and placeholder values

Decide what a missing value means for your program, then handle it before converting:

python3 - <<'PY'
def to_float(text, default=None):
    text = text.strip()
    if text in ('', 'N/A', 'null', '-'):
        return default
    return float(text)

print(to_float(''), to_float('N/A', 0.0), to_float(' 3.5 '))
PY

None 0.0 3.5

Report the bad value

When the data comes from users or files, catch the error and keep the text in your message so the row can be found:

try:
    price = float(raw)
except ValueError:
    raise SystemExit(f"invalid price in line {line_no}: {raw!r}")

Money and exactness

Binary floats cannot store most decimal fractions exactly. For amounts of money, use decimal.Decimal(text), which also raises an error (decimal.InvalidOperation) on invalid text.

Prevention

Validate input at the boundary: read CSV with explicit types, state the number format in your documentation, and reject unknown text early with a message that shows the value.

FAQ