Ran into this problem as well, appending a driver query string to the end of my connection path worked:

"mssql+pyodbc://" + uname + ":" + pword + "@" + server + "/" + dbname + "?driver=SQL+Server"

Update (July 2021) – As above, just modernized (Python 3.6+):

f"mssql+pyodbc://{uname}:{pword}@{server}:{port}/{dbname}?driver=ODBC+Driver+17+for+SQL+Server"

Note that driver= must be all lowercase.

It works using pymssql, instead of pyodbc.

Install pymssql using pip, then change your code to:

engine = sqlalchemy.create_engine("mssql+pymssql://username:password@servername/dbname")

Very late, but experienced the same such problem myself recently. Turned out it was a problem with the latest SQLAlchemy version. Had to rollback my version from 1.4.17 to 1.4.12 (unsure of in-between versions, just went with a version I knew worked).

pip install sqlalchemy==1.4.12

I had original poster's problem with a trusted connection to the Microsoft SQL Server database (pandas 1.5.3, SQLAlchemy 2.0.4). Using answers from this question, this did the trick for me:

import sqlalchemy
import pandas as pd

server = "servername"
database = "dbname"
driver = "ODBC+Driver+17+for+SQL+Server"
url = f"mssql+pyodbc://{server}/{database}?trusted_connection=yes&driver={driver}"
engine = sqlalchemy.create_engine(url)
    
query = """
    SELECT [column1]
    ,[column2] as some_other_name
    FROM [server].[dbo].[table]"""

with engine.begin() as conn:
    sqla_query = sqlalchemy.text(query)
    df = pd.read_sql(sqla_query, conn)

It should be noted that pandas is not yet fully compatible with SQLAlchemy 2.0: https://pandas.pydata.org/docs/whatsnew/v1.5.3.html